Method and apparatus for image encoding and decoding using in-loop filtering

Abstract

Provided is an image encoding method using adaptive neural network-based filtering. The method comprises the steps of: deriving a boundary strength (BS) for at least one block boundary within a reconstructed image region; determining whether to perform neural network-based filtering using the BS; and performing neural network-based filtering on the image region on the basis of the determination.

Description

Image encoding and decoding method and apparatus using in-loop filtering

[0001] The present invention relates to an image encoding / decoding method and apparatus, and more specifically, to a neural network-based in-loop filtering method and apparatus.

[0002] With the continuous development of the information and communication industry, services providing video through broadcasting and the Internet have spread globally.

[0003] Users demand videos with higher resolution and quality. To meet these user demands, video encoding and decoding technologies suitable for such videos are required. Video encoding technology can generate compressed video by compressing the video representing the images to have a smaller amount of data. Video decoding technology can generate reconstructed images using the compressed video.

[0004] Regarding video encoding and decoding technologies, various techniques exist, such as segmentation, prediction, transformation, quantization, filtering, and entropy encoding and decoding. By introducing, modifying, improving, and combining these various techniques, video and images can be compressed, transmitted, and stored more effectively.

[0005] The present invention provides a method for image encoding or decoding using adaptive neural network-based in-loop filtering.

[0006] In addition, a recording medium is provided for storing a bitstream generated by a video encoding method.

[0007] In addition, a method for transmitting a bitstream generated by a video encoding method to a video decoding device is provided.

[0008] One aspect of the present disclosure provides an image encoding method using adaptive neural network-based filtering. The method comprises: deriving a boundary strength (BS) for at least one block boundary within a restored image region; determining whether to perform neural network-based filtering using the BS; and performing neural network-based filtering on the image region based on the determination.

[0009] Another aspect of the present disclosure provides an image decoding method using adaptive neural network-based filtering. The method comprises the steps of: deriving a boundary strength (BS) for at least one block boundary within a restored image region; determining whether to perform neural network-based filtering using the BS; and performing neural network-based filtering on the image region based on the determination.

[0010] Another aspect of the present disclosure provides a method for transmitting video data to a decoder. The method comprises the steps of: encoding the video data to generate a bitstream; and transmitting the bitstream to the decoder. Herein, the step of generating the bitstream may include the step of deriving a boundary strength (BS) for at least one block boundary within a restored image region; the step of determining whether to perform neural network-based filtering using the BS; and the step of performing neural network-based filtering on the image region based on the determination.

[0011] FIG. 1 shows a system for video coding according to one embodiment.

[0012] Figure 2 shows a segmentation structure of an image according to one embodiment.

[0013] Figure 3 shows the structure of an intra prediction according to one embodiment.

[0014] FIG. 4 shows the structure of an inter prediction to explain an inter prediction process according to one embodiment.

[0015] FIG. 5 shows the order of addition of spatial candidates to the candidate list according to one embodiment.

[0016] Figure 6 shows a plurality of in-loop filters according to one example.

[0017] Figure 7 shows the structure of entropy encoding and entropy decoding according to one example.

[0018] FIG. 8 is an exemplary diagram showing the structure of an in-loop filter according to one embodiment.

[0019] FIG. 9 is an exemplary diagram showing the structure of an in-loop filter according to another embodiment.

[0020] FIG. 10 is a flowchart illustrating a neural network-based filtering process performed by an image encoding device according to one embodiment.

[0021] FIG. 11 is a flowchart illustrating a neural network-based filtering process performed by an image decoding device according to one embodiment.

[0022] Figure 12 is an example diagram showing the boundaries between blocks.

[0023] FIG. 13 is an illustrative diagram for explaining a method of setting boundary strength (BS) according to one embodiment.

[0024] FIG. 14 is an illustrative diagram for explaining a BS partition according to one embodiment.

[0025] FIG. 15 is an example diagram illustrating the number of BS samples according to one embodiment.

[0026] FIG. 16 is an example diagram illustrating the total sum of BS values ​​according to one embodiment.

[0027] FIG. 17 is an illustrative diagram for explaining sample edge size calculation according to one embodiment.

[0028] FIG. 18 is an example of samples used for calculating sample edge size according to one embodiment.

[0029] FIG. 19 is an illustrative diagram for explaining the induction of edge directionality according to one embodiment.

[0030] FIG. 20 is an example diagram showing the location where a threshold value is determined among the neural network-based in-loop filtering steps according to one embodiment.

[0031] FIGS. 21 to 23 are each illustrative diagrams for explaining whether to perform filtering on filtering blocks included in an upper region according to one embodiment.

[0032] FIG. 24 is an exemplary table showing syntactic elements that are encoded or decoded for adaptive neural network-based filtering according to one embodiment.

[0033] Various modifications may be applied to the present invention. Additionally, the present invention may have various embodiments. Specific embodiments are described by the drawings and the detailed description.

[0034] Specific embodiments are not intended to limit the invention to specific embodiments, and it should be understood that all modifications, equivalents, and substitutions that fall within the spirit and scope of the invention are included as embodiments of the invention.

[0035] The embodiments are described in sufficient detail to enable those skilled in the art to practice the embodiments. It should be understood that the various embodiments are different but need not be mutually exclusive. For example, it should be understood that the shapes, structures, and characteristics described in relation to one embodiment may be applied to or implemented in other embodiments without departing from the spirit and scope of the invention. It should also be understood that the location or arrangement of components within one embodiment may be changed without departing from the spirit and scope of the invention. Accordingly, the following detailed description is not intended to be limiting, and the scope of the exemplary embodiments is limited only by the appended claims and all equivalents to the scope claimed by such claims, provided that they are appropriately described.

[0036] The detailed description of the embodiments described below may refer to the drawings relating to the embodiments. Descriptions described in the drawings or descriptions represented by the drawings may be considered part of the detailed description. In the drawings, similar reference numerals may refer to the same or similar functions for various aspects. Dependencies between components may not be limited to those depicted in the drawings.

[0037] In the embodiments, singular expressions may include plural expressions and may be limited to and / or limited to plural expressions unless the context clearly excludes plural expressions. That is to say, in the embodiments, expressions such as 'at least one' and 'one or more' may be replaced with 'plural'. Terms such as ' / ', 'and / or', 'at least one of' and 'one or more of' described for plural items may mean 1) one of the plural items, 2) some of the plural items, 3) a combination of some of the plural items, or 4) a combination of the plural items. Additionally, plural expressions may be replaced with singular expressions. Plural may mean an integer of 1, 2, 3, 4, or 5 or more.

[0038] In the embodiments, numbered terms such as 'first' and 'second' may be used to describe various components. These terms are used solely for the purpose of distinguishing one component from another and do not limit the components. For example, without departing from the scope of the present invention, the first component may be named the second component, and similarly, the second component may be named the first component.

[0039] The statement that a first component transmits (or provides) information to a second component may mean that the first component directly transmits information to the second component, or it may mean that the first component transmits information to the second component through another third component. Here, the information received (or acquired) by the second component may be information transmitted by the first component, or information generated by applying a specific processing to information transmitted by the first component.

[0040] The components of the embodiments may be illustrated independently to represent different characteristic functions, and this does not imply that each component corresponds to a separate hardware or a single software unit. That is, the components of the embodiments may be classified and enumerated for convenience of description. Two or more components described in the embodiments may be regarded as a single component. Furthermore, a single component described in the embodiments may be separated into multiple components that perform the functions of the said component separately. Embodiments in which such components are integrated and embodiments in which components are separated are also included within the scope of the present invention, provided that they do not depart from the essence of the invention.

[0041] The terms used in the embodiments are used merely to describe specific embodiments and are not intended to limit the invention. In the embodiments, terms such as "comprising" or "having" indicate the presence of features, numbers, steps, actions, components, parts, or combinations thereof described in the embodiments. The existence or addition of other features, numbers, steps, actions, components, parts, or combinations thereof not explicitly described in the embodiments is not excluded by these terms. That is, the description of a specific component of an embodiment as "comprising" does not exclude components other than the specific component, and means that additional components may also be included within the scope of the embodiments or the technical concept of the invention.

[0042] Some of the components of the embodiments may be optional components that are not essential for performing the essential functions of the invention. Such optional components may be used to enhance performance. The embodiments may be implemented as a structure comprising only the essential components required to realize the essence of the embodiments, excluding the optional components. Such a structure is also included within the scope of the embodiments.

[0043] In the following, embodiments are described in detail with reference to the attached drawings so that a person skilled in the art can easily implement the embodiments. In describing the embodiments, if it is determined that a detailed description of related known configurations or known functions could obscure the gist of this specification, such detailed description is omitted. Additionally, the same reference numerals are used for identical components within the drawings, and redundant descriptions of identical components are omitted.

[0044]

[0045] Replacement of terms in the examples

[0046] Below, terms listed in a single line may be used with the same meaning in the examples and may be used interchangeably in the examples.

[0047] - 'one or more', 'at least one'

[0048] - 'two or more', 'a plurality of', 'multiple', 'multiple'. (In the examples, 'one or more' or 'at least one' may be further limited to 'two or more', 'multiple', or 'multiple'.)

[0049] - 'Information', 'Signal'

[0050] - 'value', 'predefined value', 'specific value', 'threshold', 'threshold value', 'baseline value', 'reference value'

[0051] - 'statistical value', 'statistics value'

[0052] - 'indicator', 'index', 'index', 'flag', 'information'

[0053] - 'encoder', 'encoding apparatus'

[0054] - 'decoder', 'decoding apparatus'

[0055] - 'Entropy encoding', 'encoding', 'encoding'

[0056] - 'Entropy decoding', 'decoding', 'decoding'

[0057] - 'Coding', 'Encoding and / or decoding'

[0058] - 'video', 'moving picture', 'image', 'picture', 'picture', 'frame', 'screen'

[0059] - 'Reference picture', 'Reference video'

[0060] - 'Reference Picture List (RPL), 'Reference Image List'

[0061] - 'original', 'input', 'source'

[0062] - 'Block', 'Unit', 'Signal'

[0063] - 'square', 'square shape'

[0064] - 'pixel', 'pixel', 'sample', 'pel'

[0065] - 'region', 'area', 'part', 'segment'

[0066] - 'partition', 'split', 'divide'

[0067] - 'quad', 'quadronary'

[0068] - 'Luma component', 'Luma', 'luminance component', 'luminance', 'Y'

[0069] - 'Chroma component', 'Chroma', 'chrominance', 'chrominance component', 'Cb and Cr', 'Cb or Cr', 'Cb', 'Cr', 'U and V', 'U or V', 'U', 'V'

[0070] - 'target', 'current' (e.g., target block and current block, or target image and current image)

[0071] - 'neighbor', 'neighboring', 'adjacent', 'neighbor / neighboring' (e.g., neighbor block, adjacent block, and neighboring block)

[0072] - 'collocated', 'COL'

[0073] - 'reconstruction', 'reconstruction', 'decoding'

[0074] - 'reconstructed', 'reconstructed', 'decoded'

[0075] - 'Difference', 'Difference', 'Difference', 'Error', 'Residual', 'Residual'

[0076] - Largest Coding Unit (LCU), Coding Tree Unit (CTU)

[0077] - 'inter', 'inter-screen'

[0078] - 'Inter prediction', 'inter prediction', 'motion compensation'

[0079] - 'Inter Mode', 'Inter Prediction Mode', 'Inter-frame Mode', 'Inter-frame Prediction Mode'

[0080] - 'Motion Vector', 'Predicted Motion Vector', 'Advanced Motion Vector Prediction (AMVP)'

[0081] - 'List', 'Candidate List'

[0082] - 'spatial candidate', 'spatial merge candidate'

[0083] - 'temporal candidate', 'temporal merge candidate'

[0084] - 'prediction motion vector candidate', 'motion vector predictor'

[0085] - 'Prediction method', 'Prediction mode'

[0086] - 'Intra', 'Inside the screen'

[0087] - 'Intra prediction', 'Intra prediction'

[0088] - 'Intra Mode', 'Intra Prediction Mode'

[0089] - 'Dequantization', 'Scaling'

[0090] - 'Quantization matrix', 'Scaling list'

[0091] - 'Quantization matrix coefficients', 'Matrix coefficients'

[0092] - 'transform coefficient level', 'quantized level', 'quantized coefficient', 'quantized transform coefficient', 'quantized transform coefficient level'

[0093] - 'dequantized coefficient', 'dequantized transform coefficient'

[0094] - 'Scanning type', 'Scanning direction'

[0095] - 'directional mode', 'angle mode', 'angular mode', 'intra-prediction mode'

[0096] - 'Intra-prediction mode (mode) number', 'Intra-prediction mode (mode) index', 'Intra-prediction mode (mode) value', 'Intra-prediction mode (mode) angle', 'Intra-prediction mode (mode) direction', 'Intra-prediction direction (mode) number', 'Intra-prediction direction (mode) index', 'Intra-prediction direction (mode) value', 'Intra-prediction direction (mode) angle'

[0097] - 'Merge Mode', 'Motion Merge Mode'

[0098] - 'Geometric Partitioning Mode (GPM)', 'Triangle Partitioning Mode'

[0099] In addition to the terms exemplified above, terms having the same meaning according to the ordinary knowledge of the technical field may be used interchangeably in the embodiments.

[0100]

[0101] Information and range of values ​​of information described in the embodiments

[0102] In the embodiments, information may include a constant, a flag, an index, a variable, a coding parameter, an element, a syntax element, motion information, an attribute, an entity, an object, and data, etc. That is to say, the term 'information' may be interchangeable with 'data', 'flag', 'index', 'variable', 'element', 'syntax element', 'motion information', 'attribute', or 'entity'.

[0103] Information can have one of multiple values. 'The nth value' can mean the nth value among multiple values.

[0104] For example, the first value can represent '0' or (logical) false. The second value can represent '1' or (logical) true. Or, the first value can represent '1' or (logical) true. The second value can represent '0' or (logical) false.

[0105] A flag may be information having a value of either '0' or '1'. In the embodiments, the values ​​'0' and '1' of the flag may be replaced with '1' and '0', respectively. For example, information indicating whether a specific process is performed or information indicating whether a specific process is applied may be considered as a flag.

[0106] When a variable such as i or j is used to represent a row, column, or index, the variable may be an integer between 0 and n - 1 inclusive. Or, the variable may be an integer between 1 and n inclusive. Here, n may be the number of rows, the number of columns, or the number of entities pointed to by the index.

[0107]

[0108] Concepts related to coding

[0109] Concepts related to coding are explained below. The descriptions disclosed below may be applied to embodiments.

[0110] Predefined value: A predefined value may refer to a value commonly used by the encoding device and the decoder. For example, a predefined value may be interpreted as being limited to a fixed value. Alternatively, a predefined value may be a value shared by the encoding device and the decoder through signaling. Alternatively, a predefined value may be a value derived through the same procedure in the encoding device and the decoder so that the encoding device and the decoder have a common value. Alternatively, a predefined value may be a common value possessed by the encoding device and the decoder. The above description of a predefined value may also apply to predefined information. In the above descriptions, 'value' may be replaced with 'information'.

[0111] - Values ​​derived through the same procedure in the above-mentioned encoding device and decoding device may include values ​​derived through the same procedure for the same value and / or the same information in the encoding device and decoding device.

[0112] - Values ​​derived through the same procedure in the above-mentioned encoding device and decoding device may include values ​​derived using the same conditional statement for the same value and / or the same information in the encoding device and decoding device.

[0113] - The description of the predefined values ​​above may also apply to predefined information. In the descriptions above, 'value' may be replaced with 'information'.

[0114] Availability: The availability of specific modes for a specific target may mean that a selected mode among the specific modes is used for that specific target. Other modes belonging to the category of specific modes may be non-available modes. Non-available modes may not be used for a specific target. The above description of specific modes may also apply to other specific information. In the above descriptions, 'mode' may be replaced with 'information'.

[0115] Adjacency: 'Direction' for 'First Object'. 'Second Object' may refer to a 'Second Object' adjacent to the 'Direction' corner / face of the First Object. For example, the 'Top-left Block' for a 'Target Block' may be a block adjacent to the top-left of the Target Block. Here, the 'First Object' may be a Target Unit, Target Block, or Target Sample. 'Direction' may be one of left-above, above, right-above, left, right, left-below, below, and right-below. The 'Second Object' may be a Unit, Block, or Sample. For the directions of top-left, top-right, bottom-left, and bottom-right, the corner of the First Object and the corner of the Second Object may be diagonally adjacent. For the directions of top, left, right, and bottom, one face of the First Object and one face of the Second Object may be in contact with each other.

[0116] - For example, the block adjacent to the top-left of the target block may be the block adjacent to the top of the block adjacent to the left of the target block. The block adjacent to the top-right of the target block may be the block adjacent to the right of the block adjacent to the top of the target block. The block adjacent to the bottom-left of the target block may be the block adjacent to the bottom of the block adjacent to the left of the target block.

[0117] Coding: Coding can refer to encoding and / or decoding of an image.

[0118] Signal: A signal can represent information about an image, unit, or block. A specific signal can represent a specific image, a specific unit, or a specific block.

[0119] Image: An image can refer to a single picture constituting a video, or it can represent the video itself. For example, "encoding and / or decoding of an image" can mean "encoding and / or decoding of a video," or it can mean "encoding and / or decoding of one of the images constituting a video."

[0120] - An image can refer to the entirety of a picture, or it can refer to a part of a picture, such as a block.

[0121] Target image: The target image may be an encoding target image that is the subject of encoding and / or a decoding target image that is the subject of decoding. Additionally, the target image may be an input image processed by an encoding device and a restored image processed by a decoding device. The target image may be an image containing a target block.

[0122] Subpicture: A picture can be divided into one or more subpictures.

[0123] - A subpicture may be a square or rectangular area within the picture. A subpicture may include one or more CTUs.

[0124] - A subpicture may include one or more slices and / or one or more tiles. For example, a subpicture may consist of one or more slice rows and one or more slice columns. Alternatively, each subpicture may consist of one or more tile rows and one or more tile columns.

[0125] - A subpicture may include one or more slices that collectively cover a rectangular area within the picture. Accordingly, the boundary of each subpicture can always be the boundary of a slice. Additionally, each vertical subpicture boundary can always be the boundary of a vertical tile.

[0126] Slice: A slice may include one or more tiles within a picture. A slice may consist of one or more rows of tiles and one or more columns of tiles.

[0127] Tile: A tile can be a square or rectangular area within a picture. A tile can contain one or more CTUs. A picture can be divided into one or more tile rows and one or more tile columns.

[0128] CTU: An image can be divided into multiple Coding Tree Units (CTUs).

[0129] - A CTU may include one Y Coding Tree Block (CTB) and at least one of a Cb CTB and a Cr CTB associated with the Y CTB, and may include information for each CTB. The information may include syntax elements.

[0130] - Each CTU may be partitioned using one or more partitioning methods to form sub-units such as Coding Units (CU), Prediction Units (PU), and Transform Units (TU). One or more partitioning methods may include Quad Tree (QT) partitioning, Binary Tree (BT) partitioning, and Ternary Tree (TT) partitioning. Additionally, each CTU may be partitioned using Multi-Type Tree (MTT) partitioning, which uses a combination of multiple partitioning methods.

[0131] CTB: CTB can refer to one of Y CTB, Cb CTB, and Cr CTB.

[0132] Unit: A unit can be determined for specific processing in coding. A unit may be information about a specific region within an image. For specific processing in coding, the image may be recursively divided into multiple parts. A unit may represent the region to which the specific processing is applied and information about the aforementioned region.

[0133] - The unit type may represent a specific process applied to the unit. Depending on the unit type, a specific process may be applied to the unit. The 'specific' unit may be a unit for the process named 'specific' in the coding. For example, the unit may be at least one of the source unit, CTU, coding unit, prediction unit, residual unit, restored residual unit, transformation unit, and restored unit.

[0134] - A unit may include samples having a two-dimensional form or arrangement. In this respect, a 'unit' may mean a 'block'. For example, a block may be at least one of an original block, a CTB, a coding block (CB), a prediction block (PB), a residual block, a restored residual block, a transform block (TB), and a restored block. For example, a partition of a unit may mean a partition of a block corresponding to the unit.

[0135] - A unit may include syntactic elements. In other words, a block and the syntactic elements for the block can be combined and referred to as a unit.

[0136] - A block is an MxN array of samples. Here, M and N can represent positive integer values, and a block can commonly represent a two-dimensional array of samples. The current block can represent the encoding target block that is the subject of encoding during encoding, or the decoding target block that is the subject of decoding during decoding. Additionally, the current block can be at least one of a coding block, a prediction block, a residual block, a transformation block, or a restoration block. Blocks can have various sizes and shapes. For example, the shape of a block can be one or more of a tetragon, a rectangular, a square, a rectangle where the width differs from the height (i.e., an oblong), a trapezoid, a triangle, a right-angled triangle, and a pentagon. Here, the width and height of the rectangle can differ from each other. Additionally, the shape of a block may include other geometric figures that can be represented in two dimensions. For example, the shape of the block may be a square or a pentagon defined by subtracting the area of ​​a right triangle from the area of ​​a rectangle. Here, the right-angled vertex of the right triangle may be one of the vertices of the rectangle. Additionally, the shape of the block may be a combination of two or more of the aforementioned shapes. Additionally, the shape of the block may be the remainder of one of the aforementioned shapes after another shape has been subtracted.

[0137] - In the embodiments, the rectangle may be limited to a non-square rectangle. When the shape of a specific object in the embodiments is described as a rectangle, this description may additionally imply that the width and height of the specific object are different from each other.

[0138] - In the embodiments, the block may be limited to at least one of a vertically oriented block and a horizontally oriented block. A vertically oriented block may mean a block in which the vertical length is greater than the horizontal length. A horizontally oriented block may mean a block in which the horizontal length is greater than the vertical length.

[0139] - The unit may include a luma component block (i.e., a Y block) and two chroma component blocks (i.e., at least one of a Cb block and a Cr block), and may include information for each block. The information may include syntax elements.

[0140] - The unit information may include the unit type, unit size, unit depth, unit encoding order, and unit decoding order.

[0141] Target Unit: The target unit may be a block that is the target of encoding, an encoding target unit, and / or a decoding target unit that is the target of decoding. The target unit may be a specific region within the target picture to which one or more specific processing steps of coding are applied. By applying a specific processing step to the target unit, a unit of a specific type may be generated. Alternatively, the target unit may represent a unit having a specific type for a specific processing step of coding.

[0142] Depth: A block can be hierarchically divided into multiple sub-blocks with depth according to a tree structure. The multiple sub-blocks created by the division of a block can be referred to as partitions.

[0143] - The block depth can represent the level of the node corresponding to the block when the blocks constituting the image are represented as a tree structure. Alternatively, the block depth can represent the number of divisions applied until the block is determined. The block depth can increase by 1 as the block is further divided.

[0144] - In a tree structure, the root node can be considered to have the smallest level, and the leaf node the largest level. The root node may be the top node of the tree structure and may correspond to the first undivided block. The level of the root node may be 0 or 1. When the level of the root node is 0, a node with level 1 may represent the block determined by the first block being divided once. A node with level n may represent the block determined by the first block being divided n times. A leaf node may be the lowest node of the tree structure. A leaf node may be a node that cannot be further divided. The depth of a leaf node may be a predefined maximum depth. For example, the maximum depth may be a positive integer such as 3. The root node may represent a CTU. A leaf node may represent at least one of CU, PU, ​​or TU.

[0145] - Depth can have a type depending on the type of partition. QT depth can represent the depth for quadtree partitioning. BT depth can represent the depth for binary partitioning. TT depth can represent the depth for ternary partitioning.

[0146] Sample: A sample can be a base unit that constitutes a block. A sample can consist of one or more bits. Bit depth can be the number of bits that make up the sample. Samples range from 0 to 2 depending on the bit depth. Bd It can be expressed as values ​​up to -1.

[0147] PU: PU may refer to a base unit for processing related to prediction. For example, processing related to prediction may include inter-prediction, intra-prediction, intra-block copy (IBC) prediction, intra-compensation, and motion compensation.

[0148] A single PU can be divided into multiple sub-PUs that are smaller in size compared to the PU. These multiple sub-PUs can also serve as base units for processing related to prediction. In other words, a prediction unit partition generated by the division of the prediction unit can also be a prediction unit.

[0149] TU: A TU may be a base unit for processing related to a residual block. Processing related to a residual block may include at least one of transform, inverse transform, quantization, inverse quantization, transform coefficient encoding, transform coefficient decoding, entropy encoding, and entropy decoding. A single TU may be divided into a plurality of sub-transform units having a size smaller than that of the TU. The plurality of sub-TUs may also be base units for processing related to a residual block. That is to say, a transform unit partition generated by the division of the transform unit may also be a transform unit.

[0150] - The transformation may include one or more of a primary transformation and a secondary transformation, and the inverse transformation may include one or more of a primary inverse transformation and a secondary inverse transformation.

[0151] Parameter set: The parameter set can correspond to header information within the structure of the bitstream.

[0152] - The parameter set may include at least one of a Video Parameter Set (VPS), a Sequence Parameter Set (SPS), a Picture Parameter Set (PPS), an Adaptation Parameter Set (APS), and a Decoding Parameter Set (DPS).

[0153] Information signaled through a parameter set can be applied to pictures that reference the parameter set. For example, information within a VPS can be applied to pictures that reference the VPS. Information within an SPS can be applied to pictures that reference the SPS. Information within a PPS can be applied to pictures that reference the PPS. A parameter set can reference a higher-level parameter set. For example, a PPS can reference an SPS. An SPS can reference a VPS.

[0154] - Additionally, the parameter set may include tile group information, slice header information, and tile header information. A tile group may refer to a group or slice containing multiple tiles.

[0155] MPM (Most Probable Mode): MPM may represent an intra prediction mode that is likely to be used for intra prediction of a target block.

[0156] - One or more different MPMs can be determined based on coding parameters related to the target block and attributes of objects related to the target block.

[0157] - One or more MPMs may be determined based on the intra prediction mode of a reference block. There may be multiple reference blocks. One or more different MPMs may be determined depending on which intra prediction modes are used for one or more reference blocks. Reference blocks may include spatial neighbor blocks.

[0158] MPM List: An MPM list may be a list containing one or more MPMs. The number of one or more MPMs in an MPM list may be predefined.

[0159] MPM Index: The MPM index can indicate one or more MPMs in the MPM list that are used for intra prediction for the target block.

[0160] MPM Usage Indicator: The MPM Usage Indicator can indicate whether an MPM list is used for prediction regarding a target block.

[0161] Prediction mode: The prediction mode may be information indicating a prediction method for a target block, such as a mode used for intra-prediction or a mode used for inter-prediction. The prediction mode may refer to one of the prediction-related modes described in the embodiments. Additionally, the prediction mode may include at least one of an intra-mode, an inter-mode, and an intra-block copy mode.

[0162] Reference image list: The reference image list may be a list containing one or more reference images used for prediction of the target block.

[0163] - There may be multiple reference image lists. Multiple reference image lists may include List 0 (List 0; L0), List 1 (List 1; L1), etc.

[0164] - One or more reference image lists may be used for inter prediction for the target block. Parts such as 'L0' and 'L1' in the names of the information related to inter prediction may refer to the reference image lists associated with the information.

[0165] Reference picture: The reference picture may be an image referenced for prediction regarding the target block. Alternatively, the reference picture may be an image containing the reference block. The reference picture may include an image prior to the target image, the target image, and an image following the target image.

[0166] Reference image index: The reference image index may be an index indicating one reference image among one or more reference images in the reference image list that is used for prediction of the target block.

[0167] Reference Block: A reference block may be a block referenced for encoding / decoding of a target block, such as for prediction and filtering. For example, a reference block may include a reference sample referenced to derive a prediction sample, and may refer to a block that provides information used for decoding the target block.

[0168] Reference Sample: A reference sample may be a sample referenced for encoding / decoding of a target block, such as prediction and filtering.

[0169] Inter prediction indicator: The inter prediction indicator may indicate the direction of inter prediction for the target block. Inter prediction may be one of unidirectional prediction and bidirectional prediction. Alternatively, the inter prediction indicator may indicate the number of reference images used when generating prediction blocks for the target block. Alternatively, the inter prediction indicator may indicate the number of prediction blocks used for inter prediction for the target block. The reference direction may refer to the inter prediction indicator. For example, the inter prediction indicator may indicate either unidirectional or bidirectional. Alternatively, for an inter mode that uses only reference images within the L0 reference image list, the inter prediction indicator may have a first value of '0'; for an inter mode that uses only reference images within the L1 reference image list, the inter prediction indicator may have a second value of '1'; and for an inter mode that uses at least two of the reference images within the L0 reference image list and the L1 reference image list, the inter prediction indicator may have a third value of '2'.

[0170] Prediction List Utilization Flag: The prediction list utilization flag for a specific reference image list may indicate whether at least one reference image within that specific reference image list is used to generate the prediction block of the target block. For example, a value of '0' for the prediction list utilization flag for a specific reference image list may indicate that the prediction block is not generated using the reference images within that specific reference image list. A value of '1' for the prediction list utilization flag for a specific reference image list may indicate that the prediction block is generated using the reference images within that specific reference image list.

[0171] - An inter-prediction indicator can be derived using prediction list utilization flags. Conversely, an inter-prediction indicator can be derived using prediction list utilization flags. For example, an inter-prediction indicator can be derived using prediction list utilization flags for multiple reference image lists. If the inter-prediction indicator indicates that specific reference lists among the multiple reference image lists are being used, the prediction list utilization flags of the specific reference lists pointed to by the inter-prediction indicator among the prediction list utilization flags of the multiple reference image lists can be set to '1', and the prediction list utilization flags of the remaining reference image lists not pointed to by the inter-prediction indicator can be set to '0'.

[0172] Reference Direction: The reference direction may point to a list of reference images used for the prediction of the target block. For example, the reference direction may point to one or more of reference image list L0 and reference image list L1. The reference direction may be used interchangeably with "inter-frame prediction direction" and may be substituted for each other.

[0173] - The reference direction merely refers to the list of reference images used for prediction of the target block, and does not indicate that the directions of the reference images within the list are restricted to a forward direction or a backward direction. That is to say, each of the reference image list L0 and the reference image list L1 may include forward images and backward images, respectively. Here, the forward direction may indicate a direction from the target image to the image preceding the target image. Forward inter-prediction may be an inter-prediction that uses the image preceding the target image as a reference image. The backward direction may indicate a direction from the target image to the image following the target image. Backward inter-prediction may be an inter-prediction that uses the image following the target image as a reference image.

[0174] - A unidirectional reference direction may mean that a single reference image list is used. A bidirectional reference direction may mean that two reference image lists are used. For example, the reference direction may indicate one of the following: that only reference image list L0 is used, that only reference image list L1 is used, or that two reference image lists are used. Additionally, the reference direction may be indicated by an inter-predictor.

[0175] Picture Order Count (POC): The POC of a picture can represent the display order or output order of the picture.

[0176] Motion information: Motion information may be information used to specify a reference block. Motion information may include information used for inter prediction, such as a motion vector (MV), reference image index, reference image, inter prediction indicator, prediction list utilization flag, etc. Additionally, motion information may include information used in a specific inter prediction mode, such as an MV candidate, MV candidate index, merge candidate, and merge index. Additionally, motion information may include information related to the block vector described below. Information related to the block vector may mean information including at least one of a block vector, a block vector candidate, and a block vector candidate index.

[0177] - Multiple motion information for multiple reference image lists may be used for inter-prediction of the target block. Motion information for a specific reference image list may be used for prediction using that specific reference image list. Multiple (intermediate) prediction blocks may be derived from the multiple motion information. A (final) prediction block for the target block may be generated using statistical values ​​for the multiple (intermediate) prediction blocks.

[0178] MV: MV can be a 2-dimensional vector used in inter-prediction. MV can represent the offset between the target block and the reference block. Alternatively, MV can represent the difference between the location of the target block and the location of the reference block.

[0179] - For example, MV is (mv x , mv y It can be expressed in the form of ). mv x can represent a horizontal component, and mv yIt can represent a vertical component.

[0180] - The zero vector can be (0, 0) MV.

[0181] Block Vector (BV): A BV can be a two-dimensional vector used in intra-block copy prediction. A BV can represent the offset between a target block within a target image and a reference block within a target image. In other words, a BV can represent the displacement between a target block and a reference block within a target image.

[0182] - For example, BV is similar to MV (bv x , bv y It can be expressed in the form of ). bv x can represent a horizontal component, and bv y It can represent a vertical component.

[0183] - The zero vector can be (0, 0) BV.

[0184] Motion Information Candidates: In a specific prediction, motion information of the target block can be selected from motion information candidates determined by a specific method. A motion information candidate may refer to the motion information of a reference block, or it may refer to the reference block itself that possesses motion information. Here, the reference block may be a block determined by a specific method to select motion information candidates.

[0185] Candidate List: A candidate list may be a list containing one or more candidates. For example, a candidate list may include a motion information candidate list, a merge candidate list, an MV candidate list, an MPM list, etc. A candidate list may be generated in the same manner in both the encoding device and the decoder. That is to say, the candidate list used in the encoding device and the candidate list used in the decoder may be identical, and the same candidate list may be shared between the encoding device and the decoder. The encoding device may select a candidate from among the candidates in the candidate list to be used for processing the target block. An indicator pointing to the selected candidate may be signaled from the encoding device to the decoder. The decoder may use the indicator to identify the candidate from among the candidates in the candidate list to be used for processing the target block. Alternatively, the encoding device and the decoder may identify the candidate from among the candidates in the candidate list to be used for processing the target block by the same rule.

[0186] Motion Information Candidate List: A motion information candidate list may refer to a list constructed using one or more motion information candidates.

[0187] Motion Information Candidate Index: The motion information candidate index may be an identifier or indicator pointing to a motion information candidate among the motion information candidates in the motion information candidate list that is used for prediction regarding the target block.

[0188] - In a specific inter-prediction mode, motion information of other restored blocks may be used to derive motion information of the target block. Other blocks may include neighboring blocks. In this specific inter-prediction mode, the motion information for the target block itself is not signaled individually, but other information used to derive motion information of the target block based on motion information of other restored blocks may be signaled. In this case, the other information may include information indicating which of the other restored blocks' motion information is used to derive motion information of the target block, such as a motion information candidate index.

[0189] - For example, these inter-prediction modes may include AMVP mode, merge mode, and skip mode. The motion information candidate index may be a merge index or an MV candidate index.

[0190] - In the embodiments, MV may be part of the motion information. In the embodiments, information about motion information, such as motion information candidates, a list of motion information candidates, and an index of motion information candidates, may be replaced with information about MV, such as MV candidates, a list of MV candidates, and an index of MV candidates, and descriptions of motion information may also be applied to MV.

[0191] Merge: Merge can refer to the merging of motion information for multiple blocks, or it can refer to applying the motion information of one block to a target block as well. In other words, merge mode can refer to a mode where the motion information of a target block is derived from the motion information of a neighboring block.

[0192] Merge Candidate: A merge candidate may refer to a specific (restored) block used for merging with a target block, or it may refer to movement information of a specific block. Alternatively, a merge candidate may include movement information of a specific block.

[0193] - Merge candidates for the target block may include spatial merge candidates, temporal merge candidates, history-based candidates, average candidates based on the average of two merge candidates, and zero merge candidates.

[0194] Merge candidate list: The merge candidate list may be a list composed of one or more merge candidates.

[0195] Merge Index: The merge index may be an indicator pointing to a merge candidate among the merge candidates in the merge candidate list that is used for prediction regarding the target block. Among the merge candidates in the merge candidate list, the movement information of the merge candidate indicated by the merge index may be used as movement information for the target block.

[0196] Neighbor block: A neighbor block may refer to a block adjacent to the target block. Neighbor blocks may include spatial and temporal neighbor blocks. A neighbor block may also refer to a reconstructed neighbor block within the reference image. A neighbor block does not necessarily have to be in direct contact with the target block.

[0197] Spatial neighbor blocks: Spatial neighbor blocks can be blocks that are spatially adjacent to the target block.

[0198] - The target block and spatial neighbor blocks can be included within the target image.

[0199] - Spatial neighbor blocks may include blocks whose boundaries, at least a portion of which abuts at least a portion of the target block's boundary. Alternatively, spatial neighbor blocks may include blocks whose distance from the target block is less than or equal to a specific value.

[0200] - Spatial neighbor blocks may include blocks diagonally adjacent to the vertices of the target block.

[0201] - Spatial neighbor blocks may include a top-left block adjacent to the top-left of the target block, a top block adjacent to the top of the target block, a top-right block entered at the top-right of the target block, a left block adjacent to the left of the target block, a right block adjacent to the right of the target block, a bottom-left block adjacent to the bottom of the target block, and a bottom-right block adjacent to the bottom-right of the target block.

[0202] Temporal neighbor blocks: Temporal neighbor blocks can be blocks that are temporally adjacent to the target block.

[0203] - Temporal neighbor blocks may include a collocated block (COL block). A collocated block may be a block within a restored image in a reference image buffer. A collocated picture (col picture) may refer to an image containing a collocated block. A collocated picture may be an image included in a reference image list.

[0204] - Call blocks can be determined based on the location of target blocks within the target image. Two blocks being 'temporarily adjacent' may mean that the locations of the two blocks satisfy certain conditions.

[0205] - The position of the call block within the call image may be the same as the position of the target block within the target image. Alternatively, the position of the call block within the call image may correspond to the position of the target block within the target image. Here, the correspondence of the block positions may mean that the regions of the blocks are identical, that the region of one block is included within the region of another block, or that one block occupies a specific location within another block.

[0206] - For example, the location of a call block within a call image may be the same as the location of a target block within a target image. Alternatively, the call block may be a block containing call samples within a call image. A call sample may be a sample having coordinates identical to the coordinates of a specific sample in the target block.

[0207] - Temporal neighbor blocks may be blocks that are temporally adjacent to the spatial neighbor blocks of the target block.

[0208] Neighbor sample: A neighbor sample may refer to a sample within a neighbor block. Neighbor samples may include prediction samples, reconstructed samples, residual samples, and decoding samples.

[0209] Search range: The search range may refer to a two-dimensional area where a search for an MV is performed during inter-prediction. For example, when an optimal MV needs to be derived for processing a target block, the optimal MV can be selected from among the MVs pointing inside the search range.

[0210] Transform coefficient: The transform coefficient may be a coefficient generated by performing a transformation on the residual block. Alternatively, the transform coefficient may be a coefficient value generated by performing inverse quantization on the quantized level.

[0211] Quantized level: A quantized level can be an integer quantity used as an input for inverse quantization.

[0212] Quantization: Quantization can be a process that generates quantized levels for transform coefficients. Quantized levels can be generated by applying quantization to transform coefficients. Transformation can also be considered as part of quantization.

[0213] Inverse Quantization: Inverse quantization can be a process of multiplying a quantized level by a factor. By applying inverse quantization to the quantized level, (restored) transformation coefficients can be generated.

[0214] Quantization Parameter (QP): QP may refer to the argument used to generate quantized levels for transform coefficients in quantization. Additionally, QP may refer to the argument used to generate (restored) transform coefficients for quantized levels in inverse quantization. Alternatively, QP may be a value mapped to the quantization step size.

[0215] Delta QP: Delta QP can be the difference between the QP predicted by a specific process and the QP of the target block. In other words, the QP of the target block can be the sum of the predicted QP and Delta QP.

[0216] Quantization matrix: A quantization matrix may be a matrix used in quantization or inverse quantization to improve the subjective or objective image quality.

[0217] Quantization matrix coefficients: Quantization matrix coefficients can be each element within the quantization matrix.

[0218] Scan: Scan can refer to a method of arranging values ​​within a block or matrix. The values ​​can be coefficients. For example, a scan can mean arranging values ​​arranged in a 2D form into a 1D form, or rearranging values ​​arranged in a 1D form into a 2D form. An inverse scan can be the opposite arrangement (or rearrangement) of the arrangement performed in a scan.

[0219] Non-zero transformation coefficients: Non-zero transformation coefficients may refer to transformation coefficients that have a non-zero value or quantized levels that have a non-zero value.

[0220] Bitstream: A bitstream may refer to a series or sequence of bits containing encoded information generated by encoding of an image. A bitstream may contain information according to specific syntax elements. For example, the information may include syntax elements. An encoding device may generate a bitstream containing information according to specific syntax elements. A decoder may obtain information from the bitstream according to specific syntax elements.

[0221] Signaling: Signaling of information may indicate that information is transmitted from an encoding device to a decoding device via a bitstream. For example, the information may include syntactic elements. Alternatively, signaling may mean that the encoding device includes information within the bitstream. Information signaled by the encoding device may be used by the decoding device. In signaling, the bitstream may be transmitted over a network and may be contained within a recording medium. In embodiments, the description that information is signaled may include: 1) the encoding device determining and generating information for signaling of information; 2) the encoding device performing encoding on the information to generate encoded information; 3) the (encoded) information being transmitted from the encoding device to the decoding device via a bitstream; 4) the decoding device performing decoding on the encoded information to obtain information; and 5) the decoding device determining and generating information through signaling of information.

[0222] - An encoding device can generate encoded information by performing encoding on the information. The encoded information can be signaled through a bitstream. A decoding device can obtain information by performing decoding on the encoded information.

[0223] - The fact that information is signaled to a specific target may mean that the information is used for each specific target, and that the processing represented by the information is applied to each specific target. For example, the fact that information is signaled at a specific unit level may indicate that the information is used or processed for each specific unit.

[0224] - The signaled information may include one or more sub-information. That specific information is signaled may mean that each piece of information of the one or more sub-information included in the specific information is signaled.

[0225] Optional Signaling: Signaling for information may be performed optionally. Optional signaling for information may mean that an encoding device optionally includes information within a bitstream (depending on specific conditions). Optional signaling for information may mean that a decoder optionally obtains information from a bitstream (depending on specific conditions).

[0226] Omission of Signaling: Signaling for information may be omitted. Omission of signaling for information may mean that the encoding device does not include information in the bitstream (depending on specific conditions). Omission of signaling for information may mean that the decoding device does not obtain information from the bitstream (depending on specific conditions). The decoding device may derive information with omitted signaling using other information of the embodiments.

[0227] Symbol: May represent at least one piece of information of a target unit, such as syntactic elements, coding parameters, quantized levels, and transform coefficients of a target unit or target block. Additionally, the symbol may represent the target of entropy encoding or the result of entropy decoding.

[0228] Entropy encoding: Entropy encoding can allocate a small number of bits to symbols with a high probability of occurrence and a large number of bits to symbols with a low probability of occurrence. Through this allocation, the size of the bitstream representing the symbols can be reduced.

[0229] Entropy coding can utilize methods such as Variable Length Coding (VLC) and Context-Adaptive Binary Arithmetic Coding (CABAC). For example, in Variable Length Coding, entropy coding can be performed using variable-length tables. For instance, in CABAC, a binaryization method for symbols and a probabilistic model of symbols / bins can be derived for entropy coding, and context-based arithmetic coding can be performed.

[0230] Entropy Decoding: In entropy decoding, the processes performed in entropy encoding can be performed in reverse. Symbols can be generated by entropy decoding of a bitstream.

[0231] Parsing: Parsing can refer to determining the values ​​of syntactic elements by performing entropy decoding on the encoded information of a bitstream. Alternatively, parsing can refer to entropy decoding itself.

[0232] Statistical Value: The values ​​of information related to specific entity(s) described in the embodiments may be used as inputs for specific operations. The statistical value may be a value derived by a specific operation on the values ​​related to these specific entity(s). For example, the statistical value for specific information may be one or more of the following: an average value, a weighted average value (weighted average), a weighted sum (weighted sum), a minimum value, a maximum value, a mode, a median value, an interpolated value, a sum of products, and a product of sums. Additionally, information of the embodiment having specific values ​​determined by operations, such as constants, variables, and coding parameters, may have a specific statistical value according to the embodiment.

[0233]

[0234] Coding parameters

[0235] In the embodiments, the coding parameters may be information required for coding. The coding parameters may include information signaled from an encoding device to a decoder, information calculated / derived during the processing of coding described in the embodiments, and information used for the processing of coding described in the embodiments.

[0236] In the embodiments, the coding parameters include the size of the CTU, the size of the unit, the form of the unit, the shape of the unit, the depth of the unit, the minimum unit size, the maximum unit size, the maximum unit depth, the minimum unit depth, the unit splitting information, QT splitting information, BT splitting information, the splitting direction of the BT splitting, the splitting form of the BT splitting, TT splitting information, the splitting direction of the TT splitting, the splitting form of the TT splitting, MTT splitting information, the combination of MTT splittings, the splitting direction of the MTT splitting, the splitting form of the MTT splitting, the prediction mode, the intra prediction mode, the luminance intra prediction mode, the chroma intra prediction mode, the intra prediction mode, the inter splitting information, the coding block splitting information, the prediction block splitting information, the transformation block splitting information, the reference sample line index, the reference sample filtering method, the reference sample filter tab, the reference sample filter coefficients, the prediction block filtering method, the prediction block filter tab, the prediction block filter coefficients, the prediction block boundary filtering method, the prediction block boundary filter tab, the prediction block boundary filter coefficients, the inter prediction mode, motion information, MV, and Motion Vector Difference; MVD), MVD resolution, MV size, MV representation accuracy, reference image list, reference image, reference image index, inter prediction direction, inter prediction indicator, prediction list utilization flag, POC, MV candidate, MV candidate index, MV candidate list, AMVP mode usage information, merge candidate, merge index, merge candidate list, merge mode usage information, motion information correction information, skip mode usage information, intra-block copy mode usage information, BV (Block Vector), Block Vector Difference (BVD), BVD resolution, BV size, BV representation accuracy, BV candidate, BV candidate index, BV candidate list, interpolation filter filter tab, interpolation filter filter coefficients, transform type, transform size, transform selection information, primary transform usage information,Secondary transform usage information, primary transform selection information, secondary transform selection information, residual block presence information, coded block pattern, coded block flag, QP, delta QP, quantization matrix, deblocking filter usage information, deblocking filter coefficients, deblocking filter filter tab, deblocking filter strength, deblocking filter shape / form, adaptive sample offset usage information, adaptive sample offset value, adaptive sample offset category, adaptive sample offset type, adaptive loop filter usage information, adaptive loop filter coefficients, adaptive loop filter filter tab, adaptive loop filter shape / form, binarization / debinarization method, context model, context model determination method, context model update method, regular mode usage information, bypass mode usage information, significant coefficient flag, last significant coefficient flag, coefficient group coding flag, last significant coefficient position, flag indicating whether the coefficient value is greater than 1, whether the coefficient value is greater than 2 Flag indicating presence, flag indicating whether the coefficient value is greater than 3, remaining coefficient value information, sign information, context bin, bypass bin, restored sample, restored luminance sample, restored chroma sample, residual sample, residual luminance sample, residual chroma sample, transform coefficient, luminance transform coefficient, chroma transform coefficient, transform coefficient level, luminance transform coefficient level, chroma transform coefficient level, transform coefficient level scanning method, quantized level, luminance quantized level, chroma quantized level, size of the MV seek area on the decoder side, shape of the MV seek area on the decoder side, number of MV seeks on the decoder side, picture type, slice identification information, slice type, slice splitting information, tile group identification information, tile group type, tile group splitting information, tile identification information, tile type, tile splitting information, bit depth,It may include one or more of input sample bit depth, restored sample bit depth, residual sample bit depth, transform factor bit depth, quantized level bit depth, mapping availability information, information about the luminance signal, information about the chroma signal, the color space of the target block, the color space of the residual block, and temporal layer information.

[0237] In addition, the coding parameter may further include 1) a value of information that may be included in the coding parameter, 2) a combination of multiple pieces of information that may be included in the coding parameter, 3) a statistical value of information that may be included in the coding parameter, 4) information related to the coding parameter, 5) information used to calculate / derive the coding parameter, and 6) information calculated / derived using the coding parameter.

[0238] In the embodiments, "X usage information" may be "information indicating whether X is used / applied / executed." Alternatively, "X usage information" may be "information indicating whether X is available." For example, "specific mode usage information" may be information indicating whether a specific mode is used. Mode information may indicate a mode used for a target block among the modes described in the embodiments. In the embodiments, specific mode usage information may be replaced with mode information, and the description of specific mode usage information may also apply to mode information. "X usage information" and "X indicator" may be used interchangeably.

[0239] In the embodiments, coding parameters and syntax elements may correspond to each other. For example, a syntax element of the embodiment may be used as a coding parameter, and a coding parameter may be signaled as a syntax element.

[0240] In the embodiments, "X existence information" may be considered as "information indicating whether X exists" or "information indicating whether information indicating X exists within the bitstream".

[0241] In the embodiments, "X selection information" may be information indicating one of the candidates or methods for X. "X selection information" may be considered as an "X index".

[0242] In the embodiments, the splitting form of a specific tree may represent one of symmetric splitting and asymmetric splitting, and may represent one of QT, BT, TT, and non-split. The splitting direction of a specific tree may represent one of horizontal direction and vertical direction.

[0243] In the embodiments, when the coding parameter has one of a plurality of values, "coding parameter" may be replaced with "whether the coding parameter has a specific value among the plurality of values ​​available to the coding parameter".

[0244] In the embodiments, when the coding parameter refers to one of a plurality of targets, the "coding parameter" may be replaced with "whether the coding parameter refers to a specific target among the plurality of targets."

[0245] In the embodiments, the coding parameter may include at least one of the type of target picture and the type of target slice. The type of target picture may be one of an I-picture, a B-picture, and a P-picture. The type of target slice may be one of an I-slice, a B-slice, and a P-slice.

[0246] - If the target image to be encoded is an I-slice, the target image can be encoded using data within the image itself without inter-predicting that references other images. For example, an I-slice can be encoded using only intra-predicting.

[0247] - If the target image is a P slice, the target image can be encoded through inter-prediction using only the reference slice existing in a unidirectional direction. Here, the unidirectional direction can be forward or reverse.

[0248] - If the target image is a B slice, the target image can be encoded through inter-prediction using reference slices existing in both directions or through inter-prediction using a reference slice existing in one of the forward and backward directions. Here, both directions can be the forward and backward directions.

[0249] - P slices and B slices encoded and / or decoded using a reference slice can be considered as images where inter-prediction is used.

[0250]

[0251] System for video coding

[0252] FIG. 1 shows a system for video coding according to one embodiment.

[0253] The system (100) may include at least one of an encoding device (110) and a decoding device (150).

[0254] Each of the encoding device (110) and the decoding device (150) may be a computer or an electronic apparatus.

[0255]

[0256] Structure of the encoding device

[0257] The encoding device (110) may include a processor (120), a storage (140), and a communicator (149).

[0258] The processor (120), storage (140), and communication device (149) can be connected via a bus.

[0259] The processor (120) may be a semiconductor device that executes instructions or computer-executable code, such as a Central Processing Unit (CPU). The processor (120) may be at least one hardware processor.

[0260] The processor (120) can perform generation and processing of information that is input to the encoding device (110) in the embodiments, output from the encoding device (110), or used inside the encoding device (110), and can perform comparison and judgment related to such information.

[0261] The processor (120) may include a plurality of components. The plurality of components may include a partitioner (122), a subtractor (124), a transformer (125), a quantizer (126), an inverse quantizer (127), an inverse transformer (128), an adder (129), a filter (130), and an entropy encoder (139).

[0262] At least some of the aforementioned multiple components may be program modules. Program modules may be included in the encoding device (110) in the form of an operating system, an application, and other program modules. Program modules may be instructions or computer-executable code stored in a storage (140) and executed by a processor (120).

[0263] The storage (140) may include various types of volatile storage media and non-volatile storage media. For example, the storage (140) may include memory such as ROM and RAM.

[0264] The storage (140) can store instructions and computer-executable code used for the operation of the encoding device (110), and can store information and bitstreams as described in the embodiments. The storage (140) may include a reference picture buffer (141).

[0265] The communication device (149) can perform functions related to the communication of information in the encoding device (110). For example, the communication device (149) can transmit a bitstream to the decoding device (150).

[0266] Among the names of the components of the encoding device (110), "-gi" ("-er" or "-or") may be replaced with "-bu" (- unit). The storage unit (140) may also be named a storage unit.

[0267]

[0268] Operation of the encoding device

[0269] The encoding device (110) can sequentially encode one or more images of the video.

[0270] The storage (140) can store the original image. In the encoding device (110), the original image can be used as the target image.

[0271] The processor (120) can generate a bitstream containing encoded information by performing encoding on the target image and can store the generated bitstream in a storage (140). The generated bitstream can be stored in a computer-readable recording medium and can be transmitted by the communication device (149) to the communication device (189) of the decoding device (150) via a wired and / or wireless transmission medium.

[0272] The splitter (122) can determine the target block by performing a split on the target image.

[0273] The predictor (123) can determine the prediction mode of the target block. The predictor (123) can generate a prediction block of the target block by performing a prediction according to the prediction mode.

[0274] The prediction mode of the target block may be one of the available prediction modes. For example, available prediction modes may include intra prediction, inter prediction, and IBC prediction.

[0275] For example, if the prediction mode is intra prediction, the predictor (123) can perform intra prediction on the target block to generate a prediction block of the target block.

[0276] For example, if the prediction mode is inter-prediction, the predictor (123) can perform inter-prediction on the target block to generate a prediction block of the target block.

[0277] For example, if the prediction mode is IBC, the predictor (123) can perform an IBC prediction for the target block to generate a prediction block of the target block.

[0278] The subtractor (124) can generate a residual block of the target block. The residual block may be the difference between the original block and the prediction block. The original block may be the region of the original image pointed to by the target block. Alternatively, the residual block may refer to a block generated by applying one or more of transformation and quantization to the difference between the original block and the prediction block.

[0279] The converter (125) can perform a conversion on the residual block to generate conversion coefficients.

[0280] The converter (125) can perform the conversion using one of a plurality of conversion methods.

[0281] For example, multiple transformation methods may include the Discrete Cosine Transform (DCT), Discrete Sine Transform (DST), Karhunen-Loeve Transform (KLT), and transformations based on each transformation.

[0282] The transform skip mode may be a mode that generates a restored block using the restored residual block and prediction block, for which transform and inverse transform have not been performed. When the transform skip mode is applied to a target block, the transform and inverse transform for the target block may be omitted, and only quantization and inverse quantization for the target block may be performed.

[0283] The quantizer (126) can generate quantized levels by applying quantization using quantization parameters to the conversion coefficients. In the embodiments, the quantized levels may also be referred to as conversion coefficients.

[0284] The entropy encoder (139) can generate encoded information by performing entropy encoding based on a probability distribution on information for decoding an image. The bitstream may contain encoded information.

[0285] Information for decoding the image may include quantized levels and syntax elements produced by the quantizer (126).

[0286] The probability distribution can be determined based on quantized levels and coding parameters.

[0287] The entropy encoder (139) can convert quantized levels, which have the form of a two-dimensional block, into the form of a one-dimensional vector by using scanning to perform encoding for the quantized levels. In the scanning, it can be determined which scan to use among an upper-right diagonal scan, a vertical scan, and a horizontal scan based on coding parameters such as the size of the block and the intra-prediction mode of the block.

[0288] When encoding is performed on a target image / block, the predictor (123) uses a reference image / block for prediction. The encoded target image / block can be used as a reference image / block for other images / blocks that are subsequently processed. Accordingly, the processor (120) can perform restoration on the encoded target block and can store the restored image containing the restored target block generated by the restoration in the reference picture buffer (141) as a reference image. Inverse quantization and inverse transform can be performed on the encoded target block for restoration.

[0289] The inverse quantizer (127) can generate inverse quantized conversion coefficients by performing inverse quantization on the quantized level.

[0290] The inverse converter (128) can generate inversely quantized and inversely converted coefficients by performing an inverse conversion on the inversely quantized conversion coefficients. In embodiments, the inversely quantized and / or inversely converted coefficients may refer to coefficients to which at least one of the inverse quantization and inverse conversion has been applied. The inversely quantized and inversely converted coefficients may be restored residual blocks.

[0291] The adder (129) can generate a recovery block by combining the prediction block and the recovered residual block.

[0292] The restoration block may pass through a filter (130). The filter (130) may apply one or more of a plurality of filters to the target. Each of the plurality of filters may be an in-loop filter. The target may be a restoration sample, a restoration block, or a restoration image.

[0293] The reference picture buffer (141) can store a restoration block / image provided from the filter (130). The restoration image may be an image containing the restoration block. Alternatively, the restoration image may be an image composed of restoration blocks.

[0294] The reference picture buffer (141) can provide the stored restored image to the predictor (123) as a reference image. In terms of storing the decoded (i.e., restored) picture, the reference picture buffer (141) may also be referred to as the Decoded Picture Buffer (DPB).

[0295]

[0296] Structure of the decoding device

[0297] The decoding device (150) may include a processor (160), a storage device (180), and a communication device (189).

[0298] The description of the processor (120), storage (140), and communication device (149) associated with the encoding device (110) may also apply to the processor (160), storage (180), and communication device (189) associated with the decoding device (150). Redundant descriptions are omitted.

[0299] The processor (160) may include a plurality of components. The plurality of components may include an entropy decoder (161), a splitter (162), a predictor (163), an inverse quantizer (167), an inverse converter (168), an adder (169), and a filter (170).

[0300] The storage (180) may include a reference picture buffer (181).

[0301] The communicator (189) can perform functions related to the communication of information in the decoding device (150). For example, the communicator (189) can receive a bitstream from the encoding device (110).

[0302] Among the names of the components of the decoding device (150), "-gi" ("-er" or "-or") may be replaced with "-bu" (- unit). The storage unit (180) may also be named a storage unit.

[0303]

[0304] Operation of the decoding device

[0305] The communication device (149) of the encoding device (110) can transmit the bitstream generated by the encoding device (100) to the decoding device (150). Alternatively, a computer-readable recording medium storing the bitstream can transmit the bitstream generated by the encoding device (100) to the decoding device (150).

[0306] The communication device (189) can receive a bitstream from the encoding device (110) via a wired and / or wireless transmission medium. The received bitstream can be stored in a storage device (180).

[0307] The processor (160) can obtain a bitstream from a storage (180) or a computer-readable recording medium.

[0308] A bitstream can contain encoded information.

[0309] The entropy decoder (161) can generate information for decoding an image by performing entropy decoding based on a probability distribution on the encoded information of the bitstream.

[0310] Information for decoding an image may include quantized levels and syntax elements, etc.

[0311] The entropy decoder (161) can convert quantized levels, which have the form of a one-dimensional vector, into the form of a two-dimensional block by using scanning to perform decoding on the quantized levels. In the scanning, it can be determined which scan to use among an upper-right diagonal scan, a vertical scan, and a horizontal scan based on coding parameters such as the size of the block and the intra-prediction mode of the block.

[0312] The entropy decoder (161) can provide syntax elements to other components of the processor (160), such as the splitter (162).

[0313]

[0314] Common explanation based on the relationship between the components of the encoding device and the components of the decoding device

[0315] The decoding device (150) performs decoding using the bitstream generated by the encoding device (110). The encoding device (110) may perform encoding for the target block using a restored image derived within the decoding device (150), rather than an original image that is not provided to the decoding device (150). Accordingly, the encoding device (110) and the decoding device (150) may need to generate the restored block / image in the same way. In this regard, the descriptions of the divider (122), predictor (123), inverse quantizer (127), inverse converter (128), adder (129), filter (130), and reference picture buffer (141) of the encoding device (110) disclosed in the embodiments may also be applied to the divider (162), predictor (163), inverse quantizer (167), inverse converter (168), adder (169), filter (170), and reference picture buffer (181) of the decoding device (150), respectively. Redundant descriptions are omitted.

[0316] Additionally, each of the divider (122), predictor (123), inverse quantizer (127), inverse converter (128), adder (129), and filter (130) of the encoding device (110) can generate syntactic element information that specifies processing for a target. Each of the divider (162), predictor (163), inverse quantizer (167), inverse converter (168), adder (169), and filter (170) of the decoding device (150) can perform processing for a target (such as that performed in the encoding device (110)) using the syntactic element information.

[0317] As described above, corresponding components of the encoding device (110) and the decoding device (150) may perform the same or corresponding functions. In embodiments, the processor may represent the processor (120) of the encoding device (110) and / or the processor (160) of the decoding device (150). For example, regarding the function of prediction, the processor may represent a predictor (123), a subtractor (124), and an adder (129), and may represent a predictor (163) and an adder (169). Regarding the function of conversion, the processor may represent a converter (125) and an inverse converter (128), and may represent an inverse converter (168). Regarding the function of quantization, the processor may represent a quantizer (126) and an inverse quantizer (127), and may represent an inverse quantizer (167). In terms of functions related to entropy encoding / decoding, the processing unit may represent an entropy encoder (139) and / or an entropy decoder (161). In terms of functions related to filtering, the processing unit may represent a filter (130) and / or a filter (170). The storage unit may represent a storage unit (140) of the encoding device (110) and / or a storage unit (180) of the decoding device (150). The reference picture buffer may represent a reference picture buffer (141) of the encoding device (110) and / or a reference picture buffer (181) of the decoding device (150). The communication unit may represent a communication unit (149) of the encoding device (110) and / or a communication unit (189) of the decoding device (150).

[0318]

[0319] Partitioning of the units that constitute the image

[0320] Figure 2 shows a segmentation structure of an image according to one embodiment.

[0321] Figure 2 schematically illustrates an example in which a single unit is divided into multiple sub-units.

[0322] CU can be used as a base unit for encoding and decoding of images. Additionally, CU can be a base unit for prediction, transformation, quantization, inverse quantization, inverse transformation, entropy encoding, and entropy decoding.

[0323] A CU can be used as a unit to which a prediction mode is applied. That is to say, in coding, it can be determined which of the available prediction modes will be applied to each CU. For example, available prediction modes may include intra prediction, inter prediction, and IBC intra block copy prediction.

[0324] The target image (200) can be sequentially divided into units of CTUs. A division structure can be determined for each CTU. The CTU can be divided into CUs according to the division structure. Alternatively, one CTU can be used as a CU. The size of the CTU can be the maximum size of the CU.

[0325] Each CU may have depth information. The depth information may represent the depth of the CU and the size of the CU. The depth of the CTU may be 0. The depth of the CU created by dividing the CTU may be 1. When a parent CU is divided into child CUs, the depth of the child CU may be 1 greater than the depth of the parent CU. The number of divided CUs may be a positive integer greater than or equal to 2, including 2, 4, 8, and 16. At least one of the width and height of the child CU created by dividing the parent CU may be smaller than at least one of the width and height of the parent CU, depending on the number of child CUs.

[0326] A partitioned CU can be recursively partitioned in the same way up to a predefined maximum depth or a predefined minimum size. The depth of a Smallest Coding Unit (SCU) can be the predefined maximum depth, and the size of an SCU can be the predefined minimum size. The size of an SCU can be the minimum CU size.

[0327] For example, the depth range of a CU can be values ​​from 0 to 3. Depending on the depth of the CU, the CU can have a size from 64x64 to 8x8. A CTU with a depth of 0 can be 64x64 blocks. 0 can be the minimum depth. An SCU with a depth of 3 can be 8x8 blocks. 3 can be the maximum depth. Depth 0 can represent a CTU that is 64x64 blocks. Depth 1 can represent a CU that is 32x32 blocks. Depth 2 can represent a CU that is 16x16 blocks. Depth 3 can represent an SCU that is 8x8 blocks.

[0328] The partition information of a CU may indicate whether the CU is partitioned. The partition information may be a 1-bit flag. All CUs except the SCU may include partition information. For example, the partition information of a CU that is not further partitioned may be a first value of '0', and the partition information of a CU that is partitioned may be a second value of '1'.

[0329] Quad Tree (QT) partitioning can mean that a single CU is partitioned into four CUs. When a parent CU is partitioned into four child CUs, the width and height of each child CU can be half the width and half the height of the parent CU, respectively.

[0330] A binary tree (BT) partition can mean that one CU is divided into two CUs. For example, if a parent CU is divided into two child CUs, the width or height of each child CU can be half the width or half the height of the parent CU.

[0331] Ternary tree (TT) partitioning can mean that a single CU is divided into three CUs. For example, when a parent CU is divided into three child CUs, the three child CUs can be created by dividing the width or height of the parent CU in a ratio of 1:2:1. The width or height of the child CUs can be 1 / 4, 1 / 2, and 1 / 4 of the width or height of the parent CU, respectively.

[0332] In FIG. 2, QT-type splitting was applied to the first CTU. QT splitting, BT splitting, and TT splitting were applied to the second CTU.

[0333] To split a CTU, at least one of different types of splits, such as QT splitting, BT splitting, and TT splitting, may be applied to the CTU. Different types of splits may be applied based on specific priorities.

[0334] For example, QT splitting may be applied preferentially to a CTU. A CU to which QT splitting can no longer be applied may correspond to a leaf node of QT. A CU that is a leaf node of QT may become a root node of BT and / or TT. A CU that is a leaf node of QT may be split into a BT form or a TT form, or may not be split further. In this case, QT splitting may not be applied again to a CU created by applying BT splitting or TT splitting to a CU that is a leaf node of QT.

[0335] The splitting of a CU corresponding to each node of QT can be signaled using QT splitting information. The QT splitting information may be a flag. The QT splitting information of a unit may be information indicating whether the unit is split into a QT form. A first value of the QT splitting information, '0', may indicate that the CU is not split into a QT form. QT splitting information having a first value may signify a Multi-Type Tree (MTT) split. MTT splitting may include BT splitting and TT splitting. A second value of the QT splitting information, '1', may indicate that the CU is split into a QT form.

[0336] There may be no priority between BT splitting and TT splitting. That is, CUs corresponding to the leaf nodes of QT can be split into BT form or TT form. Additionally, CUs generated by BT splitting or TT splitting can be split again into BT form or TT form, or they may not be split any further.

[0337] A CU corresponding to a leaf node of QT can be a root node of MTT. For a CU corresponding to each node of MTT, the CU may further include partition direction information and partition type information in the form of MTT.

[0338] The splitting direction information can indicate the splitting direction of the MTT split. The first value of the splitting direction information, '0', can indicate that the CU is split in the horizontal direction. The second value of the splitting direction information, '1', can indicate that the CU is split in the vertical direction.

[0339] The split type information may indicate the split type used for multi-type tree splitting. The first value of the split type information, '0', may indicate that CU is split into TT form. The second value of the split type information, '1', may indicate that CU is split into BT form.

[0340] Here, each of the aforementioned division direction information and division shape information may be a flag having a specified length (e.g., 1 bit).

[0341] The partitioning information of CU may also include QT partitioning information, partitioning direction information, and partitioning shape information.

[0342] A CU that is no longer divided by QT division, BT division, and TT division can be used as a unit for specific processing such as prediction, transformation, quantization, inverse quantization, inverse transformation, entropy encoding, and entropy decoding. That is, for a specific processing, the CU may no longer be divided. Therefore, division information for dividing such a CU into PU and / or TU, etc., may not exist within the bitstream.

[0343] On the other hand, if the size of a CU is larger than the maximum TU size, such a CU can be recursively partitioned until the size of the CU becomes less than or equal to the maximum TU size. For example, if the size of the CU is 64x64 and the maximum TU size is 32x32, the CU can be partitioned into 4 32x32 TUs for transformation. For example, if the size of the CU is 32x64 and the maximum TU size is 32x32, the CU can be partitioned into 2 32x32 TUs for transformation.

[0344] In such cases, information regarding whether the CU is split for transformation may not be signaled separately. Whether the CU is split may be determined without signaling by comparing the size of the CU (width / height) and the maximum TU size (width / height). For example, if the width of the CU is greater than the width of the maximum TU size, the CU may be split vertically into two. Additionally, if the height of the CU is greater than the height of the maximum TU size, the CU may be split horizontally into two.

[0345] For example, the minimum size of a CU can be 4x4. For example, the maximum size of a transformation block can be 64x64. For example, the minimum size of a transformation block can be 4x4. The minimum size of QT can be the minimum size of a CU corresponding to a leaf node of QT. The maximum depth of MTT can be the maximum depth of a path from the root node of MTT to a leaf node.

[0346] The BT maximum size may represent the maximum size of the CU corresponding to each node of the BT, and the TT maximum size may represent the maximum size of the CU corresponding to each node of the TT. The BT minimum size and / or the TT minimum size may be set as the minimum size of the CU.

[0347] If the depth of a CU within the MTT corresponding to a node of the MTT is equal to the maximum depth of the MTT, the CU may not be divided into BT form and / or TT form.

[0348] Based on the various sizes and depths of the aforementioned CU, each piece of information described in the embodiments may or may not be present in the bitstream.

[0349] Information regarding the maximum or minimum size described in the embodiments may be signaled at the upper level of the CU. In the embodiments, the upper level of the CU may include a video level, a sequence level, a picture level, a subpicture level, a tile group level, a tile level, and a slice level, etc.

[0350] The information described in the embodiments may be signaled separately for different types of slices. Different types of slices may include intra-slices and inter-slices.

[0351]

[0352] Processing of blocks based on block attributes

[0353] Whether a specific process described in the embodiments is applied or performed may be determined based on the attributes of the block associated with the specific process. Whether a specific process described in the embodiments is applied or performed may be determined based on whether the attributes of the block associated with the specific process satisfy specific conditions. For example, a block may include a target block, a neighbor block, and a reference block. A block may include other blocks described in the embodiments. A block may be one of the blocks and units described in the embodiments.

[0354] The block to which the specific treatment described in the embodiments is applied may have a square shape or a non-square shape.

[0355] In one embodiment, the attributes of the block may include the size of the block. The specific processing described in the embodiments may be applied / performed when specific conditions regarding the size of the block are met.

[0356] In one embodiment, specific conditions may include a minimum block size condition and a maximum block size condition. The block to which the minimum block size condition applies and the block to which the maximum block size condition applies may be different from each other.

[0357] In one embodiment, the minimum block size and / or maximum block size for a specific process may be predefined.

[0358] In one embodiment, the processing of the embodiment may be applied / performed when the block size is greater than or equal to the minimum block size and / or less than or equal to the maximum block size. Alternatively, in one embodiment, the processing of the embodiment may be applied / performed when the block size is greater than the minimum block size and / or less than the maximum block size.

[0359] In one embodiment, the processing of the embodiment may be applied / performed only when the block size is greater than or equal to the minimum block size and less than or equal to the maximum block size. Alternatively, the processing of the embodiment may be applied / performed only when the block size is greater than the minimum block size and less than or equal to the maximum block size. Alternatively, the processing of the embodiment may be applied / performed only when the block size is greater than the minimum block size and less than the maximum block size. The processing of the embodiment may be applied / performed only when the block size is greater than the minimum block size and less than the maximum block size.

[0360] In one embodiment, the processing of the embodiment may be applied / performed only when the block size is a predefined block size.

[0361] In the embodiments, the size of the block may be determined by various methods. For example, the size of the block may mean the width of the block or the height of the block. The size of the block may mean both the width and the height of the block. The size of the block may mean the area of ​​the block. The size of the block may mean 1) the result of a known formula using the width and height of the block, 2) the result of a formula of the embodiment, or 3) a statistical value.

[0362] Additionally, for the first size, the treatment of the first embodiment among the embodiments may be applied / performed, and for the second size, the treatment of the second embodiment among the embodiments may be applied / performed.

[0363] In the embodiments, the block size may be 2x2, 4x4, 8x8, 16x16, 32x32, 64x64, or 128x128, etc. Or, in the embodiments, the block size is (2*SIZE X )x(2*SIZE Y It may be ) etc. SIZE X is one of integers greater than or equal to 1. SIZE Y can be one of integers greater than or equal to 1.

[0364]

[0365] Predictive information for prediction

[0366] Predictive information can be used to generate a predicted block for a target block.

[0367] The encoding device (110) can generate prediction information required for prediction and can generate a bitstream containing the prediction information. The prediction information can be signaled from the encoding device (110) to the decoding device (150) through the bitstream. The decoding device (150) can obtain the prediction information from the bitstream and can generate a prediction block by performing a prediction on a target block using the prediction information.

[0368] Prediction information may include intra prediction information, inter prediction information, and IBC prediction information. In the embodiments, prediction information may be replaced with intra prediction information, inter prediction information, and / or IBC information. Intra prediction information may include information used for intra prediction as described in the embodiments. Inter prediction information may include information used for inter prediction as described in the embodiments. IBC information may include information used for IBC prediction as described in the embodiments.

[0369]

[0370] Intra prediction

[0371] Figure 3 shows the structure of an intra prediction according to one embodiment.

[0372] Intra-prediction can be performed using reference samples and coding parameters of the target block. The reference sample may be a (restored) sample within the (restored) reference block. Alternatively, an intermediate prediction sample may be generated using a sample described in an embodiment, such as the restored sample, and a reference sample may be generated again using the intermediate prediction sample. Processing described in an embodiment, such as filtering, may be applied when generating the reference sample.

[0373] The reference block may be a (spatial) neighbor block of the target block. The coding parameter may be a coding parameter for the target block and / or a coding parameter for the reference block. In intra-prediction, the reference sample may refer to a neighbor sample.

[0374] A prediction block can be generated by performing intra prediction on a target block according to an intra prediction mode, based on a reference sample within the target image and information related to the reference sample. The size of the target block and the size of the prediction block may be the same.

[0375] In the embodiments, the prediction block may be a PU. Alternatively, the prediction block may correspond to the CU or TU described in the embodiments. The prediction block may have a square or rectangular shape.

[0376] An intra prediction mode can be represented by at least one of a mode number, a mode value, a mode angle, and a mode direction. The prediction directions of a plurality of intra prediction modes for a target block are illustrated in the lower right corner of FIG. 3. Among the plurality of intra prediction modes, the remaining intra prediction modes, excluding DC and planar modes, may be directional modes. A directional mode may be an intra prediction mode having a specific direction or a specific angle. An intra prediction mode for a target block may be selected from directional modes and non-directional modes.

[0377] In the bottom-right rectangle representing the target block, the number '0' may represent Planner mode, which is a non-directional intra prediction mode. The number '1' may represent DC mode, which is a non-directional intra prediction mode. In the bottom-right rectangle representing the target block, arrows extending from the center of the rectangle outwards may represent the prediction directions of directional intra prediction modes. Additionally, the number displayed near the arrow may represent an example of a mode value assigned to an intra prediction mode or a prediction direction of an intra prediction mode.

[0378] Intra prediction can be performed according to the intra prediction mode for the target block. One of the intra prediction modes available for the target block can be used as the intra prediction mode for the target block.

[0379] The number of intra prediction modes available to the target block may be a predefined value. Alternatively, the number of intra prediction modes available to the target block may be determined based on the attributes of the prediction block. For example, the attributes of the prediction block may include coding parameters such as shape, size, and color components.

[0380] For example, in FIG. 3, the directional modes illustrated by dashed lines (i.e., directional modes with numbers from -14 to -1 or numbers from 67 to 80) can be applied only to predictions for non-square blocks. Therefore, the number of available intra-prediction modes for predictions for square blocks may be 67. (Planner mode, DC mode, and 65 directional modes)

[0381] For example, the number of available intra prediction modes may vary depending on whether the color component of a block is a luminance signal or a chroma signal. The number of available intra prediction modes for a block with a luminance component may be greater than the number of available intra prediction modes for a block with a chroma component.

[0382] Intra-prediction modes may include a horizontal-below mode, a horizontal mode, a vertical mode, and a vertical-right mode. The horizontal-below mode may be an intra-prediction mode located at the bottom of the horizontal mode. The vertical-right mode may be a mode located to the right of the vertical mode. For example, in FIG. 3, the mode value of the horizontal mode may be 18. The mode value of the vertical mode may be 50. Intra-prediction modes with a mode value of 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, and 66 may be vertical-right modes. Intra prediction modes with a mode value of 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16 and 17 may be horizontal bottom modes.

[0383] The number of the aforementioned intra-prediction modes and the mode number of each intra-prediction mode may be exemplary only. The number of the aforementioned intra-prediction modes and the mode number of each intra-prediction mode may be defined differently depending on the embodiment, implementation, and / or as necessary.

[0384] When the intra prediction mode is a planner mode, when generating a prediction block of a target block, the sample value of the prediction sample can be generated using a weighted sum (weighted sum) of the top reference sample of the target sample, the left reference sample of the target sample, the right top reference sample of the target block, and the left bottom reference sample of the target block, depending on the position of the prediction sample within the prediction block.

[0385] When the intra prediction mode is DC mode, a prediction block can be generated based on the average of the sample values ​​of multiple reference samples. The multiple reference samples may include top reference samples and left reference samples of the target block. The value of the prediction sample of the prediction block can be determined based on the average of the sample values ​​of the multiple reference samples. Additionally, filtering using the values ​​of the reference samples can be performed on specific rows and / or specific columns within the target block. The specific rows may be one or more top rows adjacent to the top reference samples. The specific columns may be one or more left columns adjacent to the left reference samples.

[0386] When the intra prediction mode is a directional mode, a prediction block can be generated using the top reference sample, left reference sample, right top reference sample, and / or left bottom reference sample of the target block.

[0387] The intra prediction mode of a target block can be determined based on the intra prediction mode of a neighboring block of the target block. Information for determining the intra prediction mode of a target block can be signaled.

[0388] For example, if the intra prediction modes of the target block and the neighbor block are the same, an indicator indicating that the intra prediction modes of the target block and the neighbor block are the same may be signaled.

[0389] For example, an indicator indicating an intra prediction mode such as the intra prediction mode of the target block among the intra prediction modes of multiple neighboring blocks may be signaled.

[0390] For example, if the intra prediction modes of the target block and neighboring blocks are different from each other, an indicator indicating the intra prediction mode of the target block may be signaled. Alternatively, information used to derive the intra prediction mode of the target block based on the intra prediction mode of the neighboring block may be signaled.

[0391] Reference samples used for intra-prediction of a target block may include bottom-left reference samples, left reference samples, top-left reference samples, top reference samples, and top-right reference samples, etc.

[0392] For example, left reference samples may be restored reference samples adjacent to the left side of the target block. Top reference samples may be restored reference samples adjacent to the top side of the target block. Top-left reference samples may be restored reference samples diagonally adjacent to the top-left side of the target block. Bottom-left reference samples may be reference samples located below the left reference samples among samples located on the same line as the left sample line composed of left reference samples. Top-right reference samples may be reference samples located to the right of the top reference samples among samples located on the same line as the top sample line composed of top reference samples.

[0393] Reference samples used for intra-prediction for a target block can be determined based on the intra-prediction mode of the target block. One or more reference samples may be used to determine the sample values ​​of the prediction samples of the prediction block. In FIG. 3, the direction of the intra-prediction mode indicated by the arrow may represent the direction from the prediction sample to the reference sample. The direction of the intra-prediction mode may represent the dependency relationship between the reference samples and the prediction samples. For example, depending on the intra-prediction mode, the sample value of a specific reference sample may be used as the sample value of at least one sample of the prediction block. Here, the specific reference sample and at least one sample of the prediction block may be samples designated by a straight line of the direction of the intra-prediction mode. That is to say, the sample value of the specific reference sample may be copied to the sample value of the prediction sample located in the reverse direction of the direction of the intra-prediction mode. Alternatively, the sample value of the prediction sample of the prediction block may be the sample value of the reference sample located in the direction of the intra-prediction mode relative to the location of the prediction sample.

[0394] Reference samples used for intra-prediction may not be limited to samples immediately adjacent to the target block. As illustrated in FIG. 3, at least one of reference sample line 0 to reference sample line 3 may be used for intra-prediction of the target block.

[0395] Each reference sample line in FIG. 3 may contain one or more reference samples. The smaller the number of the reference sample line, the closer the line of reference samples may be to the target block. Reference sample line 0 may be a line of reference samples immediately adjacent to the target block. When the top-left coordinates of the target block are (X, Y), the horizontal length is W, and the vertical length is H, the reference samples of reference sample line 0 may be samples with an x-coordinate of X-1 or a y-coordinate of Y-1. Here, the y-coordinates of the reference samples with an x-coordinate of X-1 may be Y-1 to Y+2H. The x-coordinates of the reference samples with a y-coordinate of Y-1 may be X-1 to X+2W. The reference samples of reference sample line A may be samples with an x-coordinate of XA-1 or a y-coordinate of YA-1. Here, the y-coordinates of the reference samples with an x-coordinate of XA-1 may be YA-1 to Y+2H+A. The x-coordinates of reference samples with y-coordinate YA-1 can be XA-1 to X+2W+A. A can be 1, 2, or 3.

[0396] Samples of segments A and F can be derived using padding that uses the nearest samples of segments B and E, respectively, instead of being obtained from restored neighbor blocks.

[0397] The reference sample line index may indicate a reference sample line among multiple reference sample lines used for intra-prediction of a target block. For example, the reference sample line index may have a value from 0 to 3. The reference sample line index may be signaled.

[0398] When intra-color component prediction is used for a target block, a prediction block for a second color component can be generated based on a reconstruction block of a first color component for the target block. For example, the first color component may be a luminance component, and the second color component may be a chroma component.

[0399] For intra-prediction between color components, parameters between the first and second color components can be derived based on a template. For example, the parameters can be parameters of a linear model.

[0400] For example, the template may include a top reference sample and / or a left reference sample of the target block, and may include a top reference sample and / or a left reference sample of the restoration block of the first color component corresponding to these reference samples.

[0401] Once the parameters are derived, a prediction block for a second color component for a target block can be generated by applying the reconstruction block of the first color component to a linear model. Depending on the image format or the type of intra-prediction between color components, subsampling or downsampling may be performed on the surrounding samples of the reconstruction block of the first color component and on the reconstruction block of the first color component. If subsampling is performed, the derivation of parameters and the intra-prediction between color components may be performed using corresponding samples derived by subsampling.

[0402] Intra Sub-Partitions (ISP) prediction may refer to sequential intra prediction for multiple subblocks generated by partitioning a target block. In ISP prediction, the target block may be partitioned into two or four subblocks in the horizontal and / or vertical directions. The partitioned subblocks may be restored sequentially. As intra prediction is performed on the subblocks, sub-prediction blocks for the subblocks may be generated. Additionally, as inverse quantization and / or inverse transformation is performed on the subblocks, sub-residual blocks for the subblocks may be generated. A restored subblock may be generated by adding the sub-prediction blocks to the sub-residual blocks. The restored subblocks may be used as reference samples for intra predictions for other subblocks to be processed next.

[0403] In performing a prediction for a target block, it may be determined whether samples included in a restored neighbor block can be used as reference samples for the target block. If there are non-available samples among the samples in the neighbor block that cannot be used as reference samples for the target block, a value generated by copying and / or interpolation using the sample value of at least one sample among the samples included in the restored neighbor block may replace the sample value of the non-available sample. If the value generated by copying and / or interpolation replaces the sample value of the sample, the sample may be used as a reference sample for the target block.

[0404] In intra-prediction, the sample value of a prediction sample in a prediction block can be determined by the sample value of a reference sample. The location of the reference sample can be specified by the location of the prediction sample and the direction of the intra-prediction mode. If the location specified by the location of the prediction sample and the direction of the intra-prediction mode is an integer location, the sample value of one reference sample pointed to by the integer location can be used to determine the sample value of the prediction sample in the prediction block. If the location specified by the location of the prediction sample and the direction of the intra-prediction mode is not an integer location, an interpolated reference sample can be generated based on the two reference samples closest to the specified location. The sample value of the interpolated reference sample can be used to determine the sample value of the prediction sample. That is to say, when the location specified by the location of the prediction sample and the direction of the intra-prediction mode represents the space between two reference samples, an interpolated sample value can be generated based on the sample values ​​of the two samples.

[0405]

[0406] Inter prediction

[0407] FIG. 4 shows the structure of an inter prediction to explain an inter prediction process according to one embodiment.

[0408] The rectangle shown in Fig. 4 can represent an image. Additionally, the arrow in Fig. 4 can represent the predicted direction.

[0409] Each image constituting a video can be classified into I-pictures (i.e., intra-pictures), P-pictures (i.e., uni-prediction pictures), and B-pictures (i.e., bi-prediction pictures) according to their coding type. Coding can be performed for each picture according to its coding type.

[0410] If the target picture is an I picture, coding for the target picture can be performed using information within the target picture without inter-prediction referencing other images. For example, coding for the I picture can be performed using intra-prediction and / or IBC prediction.

[0411] Coding for P picture and B picture can be performed by at least one of intra prediction, IBC prediction, and inter prediction using a reference image.

[0412] If the target picture is a P picture, coding for the target picture can be performed using unidirectional inter-prediction using a single reference image list.

[0413] When the target picture is picture B, coding for the target picture can be performed using unidirectional inter-prediction or bidirectional inter-prediction using two reference image lists.

[0414] Below, the inter prediction for the target block in the inter mode according to the embodiment is described in detail.

[0415] When the prediction mode of the target block is inter mode, inter prediction can be performed on the target block. The target block can be a prediction block or a partitioned prediction block.

[0416] Inter prediction can be performed using reference images and motion information. In inter prediction, a reference image can be selected using a reference image index, and a reference block corresponding to a target block within the reference image can be determined using motion information. A prediction block for the target block can be generated using the determined reference block.

[0417] Motion information can be derived using coding parameters, etc. For example, motion information can be derived using motion information of restored neighbor blocks, motion information of call blocks, and / or motion information of blocks adjacent to call blocks.

[0418] In the embodiments, a candidate list may be used for inter prediction. The candidate list may include multiple candidates. An index pointing to a candidate among the candidates in the candidate list that is used for inter prediction for a target block may be signaled. The candidate list may be derived in the same manner based on the same information in the encoding device (110) and the decoding device (150). Here, the same information may include a restored image and a restored block. Additionally, in order to specify a candidate by an index, the order of candidates within the candidate list may be constant.

[0419] In one embodiment, a prediction for a target block can be performed by using motion information of a spatial candidate or a temporal candidate as motion information of the target block. Motion information of a spatial candidate may be referred to as spatial motion information. Motion information of a temporal candidate may be referred to as temporal motion information.

[0420] Spatial candidates may be restored spatial neighbor blocks that are spatially adjacent to the target block.

[0421] Spatial candidates may be blocks that 1) exist within the target image, 2) have already been restored through decoding, and 3) are adjacent to the target block.

[0422] Spatial candidates may include the left block, top block, bottom-left block, top-right block, and top-left block of the target block.

[0423] Temporal candidates may be restored temporal neighbor blocks corresponding to the target block within the restored call (COL) image.

[0424] In the embodiments, the motion information of the spatial candidate may be the motion information of a block containing the spatial candidate. The motion information of the temporal candidate may be the motion information of a block containing the temporal candidate.

[0425] In inter prediction, a call block for a target block can be identified. The region of the target block within the target image and the region of the call block within the call image may be the same. That is to say, the call block may be a block that occupies a specific region within the call image. The specific region may be a region corresponding to the region of the target block within the call image.

[0426] Temporal candidates may be locations inside and / or outside the call block within the call image.

[0427] For example, a call block may include a first call block and a second call block. When the top-left coordinates of the call block are (xP, yP) and the size of the call block is (nPSW, nPSH), the first call block may be a block occupying the coordinates (xP + nPSW, yP + nPSH). The second call block may be a block occupying the coordinates (xP + (nPSW >> 1), yP + (nPSH >> 1)). The second call block may optionally be used as a call block when the first call block is unavailable.

[0428] The MV of the target block can be determined based on the MV of the call block. Scaling can be performed on the MV of the call block. The scaled MV of the call block can be used as the MV of the target block or the prediction MV. Alternatively, the MV of the temporal candidate stored in the candidate list associated with the inter-prediction can be the scaled MV.

[0429] The ratio of the scaled MV and the MV of the call block may be equal to the ratio of the first temporal distance and the second temporal distance. The first temporal distance may be the distance between the reference image of the target block and the target image. The second temporal distance may be the distance between the reference image of the call block and the call image.

[0430] The method by which motion information is derived can be determined by the inter prediction mode of the target block. For example, as an inter prediction mode, AMVP mode, merge mode, skip mode, merge mode with MVD, sub-block merge mode, GPM, Combined Inter Intra Prediction (CIIP) mode, and affine inter mode may be used. In the following embodiments, each of the inter prediction modes is described.

[0431]

[0432] AMVP mode

[0433] When AMVP mode is used as a prediction mode, a list of MV candidates including one or more MV candidates can be generated using spatial candidate MVs, temporal candidate MVs, history-based MV candidates, and zero vectors. At least one of the spatial candidate MVs, temporal candidate MVs, and zero vectors can be determined and used as an MV candidate.

[0434] Spatial candidates may include restored spatial neighbor blocks. The MV of a restored spatial neighbor block may be referred to as a spatial motion vector candidate. Temporal candidates may include a call block and a block adjacent to the call block. The MV of a call block or the MV of a block adjacent to the call block may be referred to as a temporal motion vector candidate. History-based MV candidates may be MVs in a list containing MVs of other blocks that were encoded / decoded before the encoding / decoding of the target block.

[0435] The encoding device (110) can determine the MV to be used for encoding a target block within a search range using an MV candidate list. The maximum number of MV candidates in the MV candidate list may be predefined. N may represent the predefined maximum number. For example, N may be 2. Alternatively, the maximum number of such candidates may be signaled from the encoding device to the decoding device or derived from the decoding device. The encoding device (110) can determine an MV candidate to be used as the predicted MV of the target block among the MV candidates in the MV candidate list. The MV to be used for encoding the target block may be an MV that can be encoded at the minimum cost. The encoding device (110) may determine whether to use the AMVP mode in encoding the target block and may generate AMVP mode usage information indicating whether the AMVP mode is used.

[0436] Inter prediction information may include 1) AMVP mode usage information, 2) MV candidate index, 3) MVD, 4) MVD resolution information, 5) reference direction and 6) reference image index, and may include residual blocks. Inter prediction information may be signaled from the encoding device (110) to the decoding device (150) in the form of a bitstream.

[0437] The decoding device (150) can obtain AMVP mode usage information from the bitstream. If the AMVP mode usage information indicates that the AMVP mode is being used, the decoding device (150) can obtain an MV candidate index, an MVD, MVD resolution information, a reference direction, and a reference image index from the bitstream. Among the MV candidates included in the MV candidate list, the MV candidate pointed to by the MV candidate index can be selected as the predicted MV of the target block.

[0438] The MVD may represent the difference between the MV that will actually be used for inter-prediction of the target block and the predicted MV. The encoding device (110) may derive a predicted MV that is close to the MV that will actually be used for inter-prediction of the target block in order to use an MVD of the smallest possible size. The decoding device (150) may derive the MV of the target block by summing the MVD and the predicted MV. That is to say, the MV of the target block derived by the decoding device (150) may be the sum of the MVD and the predicted MV candidates.

[0439] Additionally, the encoding device (110) can generate MVD resolution information. The MVD resolution information may be information used to adjust the resolution of the MVD. The decoding device (150) can adjust the resolution of the MVD using the MVD resolution information.

[0440] Meanwhile, the encoding device (110) can calculate the MVD based on an affine model. The affine control point MV of the target block can be derived based on the sum of the affine control point MV candidates and the MVD. Using the affine control point MV, the MV of each sub-block within the target block can be derived.

[0441]

[0442] Merge Mode

[0443] When merge mode is used, a merge candidate list containing multiple merge candidates can be generated using motion information of spatial candidates and motion information of temporal candidates, etc. Motion information may include 1) MV, 2) reference image index and 3) reference direction, etc. A merge candidate may be motion information.

[0444] Merge candidates may include 1) spatial merge candidates generated based on spatial candidates, 2) temporal merge candidates generated based on temporal candidates, 3) history-based merge candidates, 4) average merge candidates, and 5) zero merge candidates.

[0445] A history-based merge candidate may be movement information within a list containing movement information of other blocks that were encoded / decoded earlier than the encoding / decoding of the target block.

[0446] The average merge candidate may be a merge candidate generated based on the average of two merge candidates within the merge candidate list.

[0447] Zero merge candidates can be zero vector motion information. Zero vector motion information can be motion information where MV is a zero vector.

[0448] Merge candidates can be added to the merge candidate list according to a predefined method and a predefined order so that the merge candidate list has a set number of merge candidates. The same merge candidate list can be configured in the encoding device (110) and the decoding device (150) through the predefined method and a predefined order.

[0449] The encoding device (110) can select a merge candidate to be used for encoding a target block from among the merge candidates in the merge candidate list. The encoding device (110) can determine whether to use a merge mode in encoding the target block and can generate merge mode usage information indicating whether the merge mode is used.

[0450] Inter prediction information may include 1) merge mode usage information, 2) merge index and 3) correction information, etc., and may include residual blocks. Inter prediction information may be signaled in bitstream form from the encoding device (110) to the decoding device (150).

[0451] The decoding device (150) can obtain merge mode usage information from the bitstream. If the merge mode usage information indicates that the merge mode is being used, the decoding device (150) can obtain merge mode-related information, such as a merge index, from the bitstream.

[0452] The encoding device (110) can select the optimal merge candidate among the merge candidates included in the merge candidate list and can set the value of the merge index to point to the selected merge candidate.

[0453] Correction information may be information used for correcting the MV. The encoding device (110) may generate correction information. The decoding device (150) may derive a corrected MV by performing correction on the MV of a merge candidate selected by a merge index based on the correction information. The corrected MV may be used as the MV of the target block.

[0454] In one embodiment, the correction information may include an MVD. The correction information may include one or more of correction usage information, correction direction information, and correction magnitude information. The correction usage information may indicate whether to use correction for the MV. A merge mode that performs correction for the MV based on the correction information may be referred to as a merge mode having an MVD.

[0455] In merge mode, a prediction for the target block can be performed using the merge candidate pointed to by the merge index among the merge candidates included in the merge candidate list.

[0456] Movement information of the target block can be derived from 1) MV, 2) reference image index and 3) reference direction of the merge candidate pointed to by the merge index.

[0457] In one embodiment, the merge candidates in the merge candidate list may be specific modes that induce inter-prediction information. A merge candidate may be information pointing to a specific mode that induces inter-prediction information. Inter-prediction information of a target block may be induced according to the specific mode pointed to by the merge candidate. In this regard, the specific mode may be regarded as a specific inter-prediction information inducing mode or a specific movement information inducing mode. The specific mode may include a series of processes that induce inter-prediction information.

[0458] Inter-prediction information of the target block can be derived according to a specific mode pointed to by a merge candidate selected by a merge index among the merge candidates in the merge candidate list. For example, specific modes may include a mode for deriving motion information at the sub-block level and a mode for deriving motion information at the affine level, and may include other modes for deriving motion information as described in the embodiments.

[0459] Skip mode may be a mode that does not use residual blocks. That is to say, when skip mode is used, the restoration block may be identical to the prediction block. The description of the merge mode in the embodiments may also apply to skip mode. The difference between merge mode and skip mode may be whether or not residual blocks are signaled and used. That is to say, skip mode may be similar to merge mode except that residual blocks are not transmitted / used, and the description of merge mode may also apply to skip mode.

[0460] The subblock merge mode may be a mode in which motion information of a target subblock is induced for a target subblock within a target block. When the subblock merge mode is applied, a list of subblock merge candidates may be generated using affine control point motion vector merge candidates and / or subblock-based temporal merge candidates. The subblock-based temporal merge candidates may be motion information of the call subblock of the target subblock.

[0461] In GPM, a first prediction block and a second prediction block can be generated using two sets of motion information for a target block. For each coordinate of the target block, a final prediction sample of the final prediction block can be generated using the weighted sum of the first prediction sample of the first prediction block and the second prediction sample of the second prediction block.

[0462] Here, the first weight for the first prediction sample of the weighted consensus and the second weight for the second prediction sample can be determined based on the boundaries of the GPM. The boundaries may represent dividing lines that divide the target block. Depending on the boundaries, the target block may be divided into a first divided region and a second divided region.

[0463] If the distance between the final prediction sample and the boundary is less than or equal to a reference value, the value of the final prediction sample of the final prediction block may be determined using the weighted sum of the first prediction sample of the first prediction block and the second prediction sample of the second prediction block. If the distance between the final prediction sample and the boundary is greater than the reference value, one of the first weight and the second weight may be 1 and the other may be 0.

[0464] The Combined Inter-Intra Prediction (CIIP) mode may be a mode that derives a prediction sample of a target block using a weighted sum of a prediction sample generated by inter-prediction and a prediction sample generated by intra-prediction.

[0465] In the aforementioned modes, self-improvement of the derived motion information may be performed, and the improved motion information may be used as motion information for the target block. For example, blocks within a specific area determined based on the derived motion information may be searched, and the motion information of the block having the smallest Sum of Absolute Differences (SAD) value among the searched blocks may be used as the improved motion information for the target block. The specific area may be a square area within the reference image specified by the motion information. The point indicated by the motion information may be the center of the specific area.

[0466] In the aforementioned modes, compensation for prediction samples derived through inter-prediction can be performed using optical flow.

[0467]

[0468] FIG. 5 shows the order of addition of spatial candidates to the candidate list according to one embodiment.

[0469] In Fig. 5, the locations of the spatial candidates are shown.

[0470] The large block in the center can represent the target block. The five small blocks adjacent to the target block can represent spatial candidates.

[0471] The coordinates of the target block can be (xP, yP), and the size of the target block can be (nPSW, nPSH).

[0472] Spatial candidate A0 may be a block adjacent to the bottom-left of the target block. A0 may be a block occupying a sample of coordinates (xP - 1, yP + nPSH).

[0473] Spatial candidate A1 may be a block adjacent to the left of the target block. A1 may be the bottommost block among the blocks adjacent to the left of the target block. Or, A1 may be a block adjacent to the top of A0. A1 may be a block occupying a sample of coordinates (xP - 1, yP + nPSH - 1).

[0474] Spatial candidate B0 may be a block adjacent to the top right of the target block. B0 may be a block occupying a sample of coordinates (xP + nPSW, yP - 1).

[0475] Spatial candidate B1 may be a block adjacent to the top of the target block. B1 may be the rightmost block among the blocks adjacent to the top of the target block. Or, B1 may be a block adjacent to the left of B0. B1 may be a block occupying a sample of coordinates (xP + nPSW - 1, yP - 1).

[0476] Spatial candidate B2 may be a block adjacent to the top-left corner of the target block. B2 may be a block occupying a sample of coordinates (xP - 1, yP - 1).

[0477] As illustrated in Fig. 5, in adding spatial candidates to the candidate list, B1, A1, The order B0, A0, and B2 can be used. That is, B1, A1, Available spatial candidates can be added to the candidate list in the order of B0, A0, and B2. The order in which spatial candidates illustrated in FIG. 5 are added to the merge candidate list may be just one example.

[0478] The above candidate list may include a motion information candidate list, a merge candidate list, an MV candidate list, a BV candidate list, and an MPM list, etc.

[0479] To include a spatial or temporal candidate in the candidate list, it may be determined whether the spatial or temporal candidate is available. If a candidate block is outside the boundaries of an image, slice, or tile, the availability of the candidate block may be set to false. The description "availability is set to false" may mean "it is set to non-available."

[0480] The maximum number of candidates in the candidate list can be set. N can represent the set maximum number. The set maximum number can be signaled through a parameter set or header, etc. For example, the maximum number of candidates in the candidate list for a target block within a slice can be set by the slice header. For example, the value of N can be 5 by default.

[0481]

[0482] IBC mode

[0483] The IBC mode may be an intra-block copy prediction mode that generates a prediction block for a target block by referencing an already restored region within the target image. In this respect, the IBC mode may also be referred to as a current image reference mode. A block vector (BV) may be used to identify the already restored region.

[0484] Whether the target block is encoded / decoded in IBC mode can be determined using IBC mode usage information. The encoding device (110) can determine whether to use IBC mode in encoding the target block and can generate IBC mode usage information indicating whether IBC mode is used. The decoding device (150) can obtain IBC mode usage information from the bitstream.

[0485] In IBC mode, the predicted block of the target block can be generated based on the BV. The BV can specify the reference block. The BV can indicate the displacement between the target block and the reference block. The reference block can be a block within the target image. The description of the MV of the embodiments can also be applied to the BV.

[0486] The IBC mode may include a skip mode, a merge mode, and an AMVP mode, etc. The descriptions of the AMVP mode, merge mode, and skip mode of the embodiments may be similarly applied to the AMVP mode, merge mode, and skip mode of the IBC mode, respectively.

[0487] In skip mode or merge mode, a merge candidate list may be configured, and a merge index may specify one merge candidate from among the merge candidates in the merge candidate list. The BV of the specified merge candidate may be used as the BV of the target block.

[0488] In AMVP mode, BVD can be used. The description of MVD in the embodiments can also be applied to BVD.

[0489] The reference block in IBC mode may be limited to a block within an already restored region of the target image. Alternatively, the reference block may be contained within at least one of the target CTU or the left CTUs. For example, the value of BV may be restricted so that the reference block is located within a specific region. The specific region may be an area of ​​three blocks of a specific size that are encoded / decoded before the block of a specific size containing the target block. The specific size may be 64x64.

[0490]

[0491] Transformation and Quantization

[0492] Quantized levels can be generated by performing a transformation and / or quantization on the residual block. The residual block can represent the difference between the original block and the prediction block. A restored residual block can be generated by performing inverse quantization and / or inverse transformation on the quantized levels. The restored residual block can represent the difference between the restored block and the prediction block.

[0493] When a transformation or inverse transformation is performed, a separable transform or a 2D non-separable transform may be performed on the residual block. A separable transform may be a transformation that performs 1D transformations on the residual block in the horizontal and vertical directions, respectively.

[0494] The transformation kernels used for the transformation may include various DCT kernels such as DCT type 2 (DCT-II), 2) DST kernels, and 3) kernels derived by training. For 1D transformation, DCT type and DST type may include DCT-V, DCT-VIII, DST-I, and DST-VII in addition to DCT-II.

[0495] A transformation set may be used to determine the DCT type, DST type, or learning-derived kernel to be used for the transformation. Each transformation set may include multiple transformation candidates. Each transformation candidate may be a DCT type, a DST type, or a learning-derived kernel, etc.

[0496] The encoding device (110) can perform conversion and inverse conversion using conversion candidates included in the conversion set. The decoding device (150) can perform inverse conversion using conversion candidates included in the conversion set. Conversion selection information indicating which conversion candidate is used among the plurality of conversion candidates included in the conversion set applied to the residual block may be signaled. The conversion selection information may include vertical conversion selection information and horizontal conversion selection information. The vertical conversion selection information may indicate which conversion among the conversions belonging to the conversion set is used for the vertical conversion. The horizontal conversion selection information may indicate which conversion among the conversions belonging to the conversion set is used for the horizontal conversion.

[0497] The transformation may include at least one of a primary transformation and a secondary transformation. A primary transformation coefficient may be generated by performing a primary transformation on a residual block, and a secondary transformation coefficient may be generated by performing a secondary transformation on the transformation coefficient. Here, the transformation coefficient may include a primary transformation coefficient and a secondary transformation coefficient.

[0498] A first-order transformation may mean a Multiple Transform Selection (MTS) that applies different transformations to each of the 1D directions (i.e., vertical and horizontal directions).

[0499] A second-order transformation may be a transformation intended to improve the energy concentration of the transformation factors generated by a first-order transformation. A second-order transformation may be 1) a separable transformation like the first-order transformation, or 2) a 2D non-separable transformation. A 2D non-separable transformation may refer to a Low Frequency Non-Separable Transform (LFNST) or a Non-Separable Primary Transform (NSPT).

[0500] NSPT can be applied to specific block sizes such as 4x4, 4x8, 8x4, 4x16, 16x4, 8x8, 8x16, and 16x8 for intra-coding.

[0501] A first-order transformation may be performed using at least one of a plurality of predefined transformation methods. For example, the plurality of predefined transformation methods may include DCT, DST, and KLT, etc. Additionally, the first-order transformation may be a transformation having various transformation types according to transformation kernel functions that define DCT and DST. For example, the first-order transformation may include a plurality of transformations such as DCT-2, DCT-4, DCT-5, DCT-7, DCT-8, DST-1, DST-2, DST-4, DST-7, and DST-8 according to a plurality of transformation kernels.

[0502] In one embodiment, the transformation type may be determined based on coding parameters related to the target block. For example, the transformation type may be determined based on one or more of 1) the prediction mode of the target block (e.g., one of intra prediction and inter prediction), 2) the size of the target block, 3) the shape of the target block, 4) the intra prediction mode of the target block, 5) the components of the target block (e.g., one of luminance components and chroma components), and 6) the splitting type applied to the target block (e.g., one of QT, BT, TT, and non-split).

[0503] As in the first transformation, a set of transformations can be defined in the second transformation as well. Methods for deriving and / or determining the set of transformations of the embodiments can be applied to the second transformation as well as the first transformation.

[0504] In one embodiment, the first transformation and / or second transformation may be determined for a specific target. The transformation selection information may include transformation target information. The transformation target information may indicate the target to which the first transformation and / or second transformation is applied.

[0505] For example, first-order transformation and / or second-order transformation may be applied to one or more signal components among the luminance component and the chroma component.

[0506] In one embodiment, the transformation selection information may include first transformation usage information and second transformation usage information. The first transformation usage information may indicate whether a first transformation is applied to the residual block of the target block. The second transformation usage information may indicate whether a second transformation is applied to the residual block of the target block.

[0507] In one embodiment, whether a first transformation and / or a second transformation is applied may be determined based on coding parameters for the target / neighbor block, such as the size and shape of the target / neighbor block.

[0508] In one embodiment, the transformation selection information may include first transformation selection information and second transformation selection information. The first transformation selection information may indicate a transformation method applied to a residual block among a plurality of transformation methods that can be used in the first transformation. The first transformation selection information may be a first transformation index. The second transformation selection information may indicate a transformation method applied to a transformation coefficient among a plurality of transformation methods that can be used in the second transformation. The second transformation selection information may be a second transformation index.

[0509] In one embodiment, the transformation methods of the first transformation and the second transformation can each be derived based on specific information such as coding parameters. For example, the coding parameters may include coding parameters for target / neighbor blocks.

[0510] In the embodiments, information related to a transformation, such as transformation selection information, and sub-information of the transformation selection information may be signaled to a specific target. For example, the specific target may be a CU.

[0511] Information related to transformations, such as transformation selection information, and sub-information of transformation selection information can be derived for a specific target. For example, the specific target may be a CU.

[0512] Quantized levels can be generated by performing quantization on the result or residual block generated by performing a first-order transformation and / or a second-order transformation.

[0513] The description of the transformation described above may also be applied to the inverse transformation. In such application, the inverse processing of the processing described for the transformation may be performed in the inverse transformation. "Transformation" within the name related to the transformation may be changed to "inverse transformation." Additionally, the input of the transformation may be considered as the output of the inverse transformation. The output of the transformation may be considered as the input of the inverse transformation. The decoding device (150) may obtain information related to the transformation, such as transformation selection information, and may use the information related to the transformation to perform the inverse processing of the transformation related to the transformation indicated by the information related to the transformation.

[0514] The target block may include multiple subblocks. Each subblock may be defined according to a minimum block size or minimum block shape. The target block may be divided into multiple subblocks, and each subblock may include coefficients such as 4x4, 2x8, and 8x2. The target block may be a transformation block. Transform coefficients or quantized levels may be represented in the form of a block. Transform coefficients may be quantized transformation coefficients.

[0515] Transform coefficients or quantized levels may be scanned according to at least one scanning type among diagonal scanning, vertical scanning, and horizontal scanning. Diagonal scanning may be top-right diagonal scanning or bottom-left diagonal scanning.

[0516] For example, by scanning the coefficients of a block using diagonal scanning, the coefficients can be changed or arranged into a one-dimensional vector form. Vertical scanning may be scanning the coefficients in the form of a two-dimensional block in a column direction. Horizontal scanning may be scanning the coefficients in the form of a two-dimensional block in a row direction.

[0517] The scanning type for the coefficients can be determined based on coding parameters such as intra prediction mode, block size, and block shape. For example, based on coding parameters such as intra prediction mode, block size, and block shape, it can be determined which scanning method—diagonal scanning, vertical scanning, and horizontal scanning—will be used. A block may be a transformation unit.

[0518] Scanning according to each scanning type can start at a specific starting point and end at a specific ending point.

[0519] In scanning, the scanning order according to the scanning type can first be applied between subblocks. Next, the scanning order according to the scanning type can be applied to the transformation coefficients or quantized levels within the subblocks.

[0520] The encoding device (110) can perform entropy encoding on the conversion coefficients or quantized levels to generate a bitstream containing entropy-encoded conversion coefficients or entropy-encoded quantized levels.

[0521] The decoding device (150) can generate the transform coefficients or quantized levels by obtaining entropy-encoded transform coefficients or entropy-encoded quantized levels from the bitstream and performing entropy decoding. The coefficients can be arranged in the form of two-dimensional blocks through inverse scanning. The arrangement of inverse scanning may be a rearrangement opposite to the arrangement of scanning.

[0522] Backscanned transform coefficients or backscanned quantized levels can be generated through backscanning of the coefficients. In this case, the backscanning types of backscanning may include diagonal scans, vertical scans, and horizontal scans, and a backscanning type of the inverse transform corresponding to the scanning type of the transform may be selected.

[0523] In the decoding device (150), inverse quantization can be performed on the (backscanned) coefficients. Depending on whether a second inverse transform is performed, a second inverse transform can be performed on the result generated by the performance of inverse quantization. Also, depending on whether a first inverse transform is performed, a first inverse transform can be performed on the result generated by the performance of the second inverse transform. By selectively performing a second inverse transform and a first inverse transform on the coefficients, a restored residual block can be generated.

[0524]

[0525] Filtering

[0526] To improve the image quality, filtering may be performed on the blocks. The value of the target sample may be determined or updated by the filtering.

[0527] The target sample may be one of the samples described in the embodiments. For example, the target sample may be one or more of the samples described in the embodiments, such as a prediction sample, a reference sample, a residual sample, a reconstructed sample, and a reconstructed sample to which filtering has been applied.

[0528] The target sample may be a sample within one or more of the target picture, target slice, target CTB, target block, reference sample line, and template. The target block may be one of the blocks described in the embodiments. For example, the target block may be one or more of the blocks described in the embodiments, such as a transformation block, prediction block, reference block, residual block, and restoration block.

[0529] In the embodiments, the filtering process described as being applied to one target may also be applied to other targets. For example, the filtering process described in a specific in-loop filtering may also be applied to a transformation block, a prediction block, a reference block, and a residual block, etc.

[0530] For the filtering of the embodiments, a specific type of filtering may be used. The type of filtering may include a filter tap (or filter tap length), a filter shape, a filter strength, filter coefficients (or weights), and an offset.

[0531] The filter tab may indicate the number of input samples used for the filter. The input samples may include target samples. Alternatively, the input samples may include specific values ​​determined for the target samples. The input samples may include one or more reference samples. One or more reference samples may be determined based on the attributes of the target block described in the embodiments. The attributes may include coding parameters. For example, the attributes of the target sample may include the location of the target sample. One or more reference samples may be specified based on their relative position to the location of the target sample.

[0532] The filter shape can represent the shape formed by input samples. A specific value determined for a target sample can be considered as the target sample. In other words, if a specific value determined for a target sample is used as an input sample for a filter, the target sample can also be considered as constituting the filter shape.

[0533] There may be multiple samples whose values ​​are determined by filtering. Filter strength may represent the range of samples whose values ​​are determined by filtering. Filter strength may be either strong filtering strength or weak filtering strength. The number of samples whose values ​​are determined by strong filtering strength may be greater than the number of samples whose values ​​are determined by weak filtering strength. Alternatively, filter strength may represent the range of values ​​that are changed by filtering. The range of sample values ​​changed by strong filtering strength may be wider than the range of sample values ​​changed by weak filtering strength.

[0534] Filter coefficients can be coefficients or weights of the input samples.

[0535] The offset can be a specific value added to the result calculated using the values ​​and coefficients of the input samples, such as a weighted sum.

[0536] Filtering, interpolation, and sampling may be common in that they update the values ​​of samples. Accordingly, the description of any one of filtering, interpolation, and sampling in the embodiments may also apply to the other one of filtering, interpolation, and sampling. Here, sampling may include at least one of upsampling, downsampling, and subsampling.

[0537] Filtering may include filtering performed by a predictor (123) and a predictor (163), etc.

[0538] In encoding for a target block, a prediction error may exist between the original sample of the original block and the prediction sample of the prediction block. To reduce the prediction error, filtering may be performed on at least one of the prediction sample of the prediction block and the reference sample referenced for prediction.

[0539] For example, in intra-prediction, the reference samples may include one or more of the top-left reference sample, top reference sample, top-right reference sample, left reference sample, and bottom-left reference sample. Filtering of the prediction samples may be performed by applying specific weights to the prediction samples, left reference samples, top reference samples, and / or top-left reference samples, respectively.

[0540] Filtering for at least one of the prediction sample and the reference sample may be performed based on the attributes of the target block and the attributes of the prediction sample. For example, whether filtering is performed, the type of filter, the area to which filtering is applied, the weights of the filtering, the reference sample, the range of the reference sample, and the location of the reference sample may each be determined based on the attributes of the target block and the attributes of the prediction sample.

[0541] For example, the attributes of the target block may include information related to the target block described in the embodiments, such as 1) size, 2) prediction mode, 3) intra prediction mode, 4) reference sample line, 5) sample value, and 6) coding parameter.

[0542] For example, the attributes of the prediction sample may include information related to the prediction sample described in the embodiments, such as 1) the sample value and 2) the location within the target block of the prediction sample, and may include coding parameters regarding the prediction sample.

[0543] Filtering may include in-loop filtering performed by a filter (130) and a filter (170), etc.

[0544]

[0545] Figure 6 shows a plurality of in-loop filters according to one example.

[0546] Multiple in-loop filters of in-loop filtering may include one or more of Luma Mapping with Chroma Scaling (LMCS), deblocking filter, Sample Adaptive Offset (SAO), and Adaptive Loop Filter (ALF).

[0547] Multiple in-loop filters can be connected sequentially. For example, multiple in-loop filters can be connected in the order of LMCS, deblocking filter, SAO, and ALF. Additionally, multiple in-loop filters can be connected in any order of all available permutations of the multiple in-loop filters. The output from one of the multiple in-loop filters can be used as an input to the next filter.

[0548] As illustrated in FIG. 6, an input image may be input to the first filter. The input image may be a block as described in the embodiments. For example, the input image may be a restored block generated by an adder (129) or an adder (169). The output from one filter may be input to the next filter. An output image may be generated by the last filter. The output image may be a filtered block as described in the embodiments. For example, the output image may be a filtered restored image generated by a filter (130) or a filter (170).

[0549] The target block can represent the image input to the filter. The filtered target block can represent the image output from the filter.

[0550] LMCS may include luminance signal mapping for the luminance signal of the target block and chroma signal scaling for the chroma signal of the target block.

[0551] Luminous signal mapping can perform codeword redistribution for the luminous signal.

[0552] Luma signal mapping may include forward mapping and inverse mapping. In forward mapping, the existing dynamic range may be divided into multiple intervals. The mapped dynamic range can be determined by performing codeword redistribution on the input image using a linear model for each interval. In inverse mapping, inverse mapping from the mapped dynamic range to the existing dynamic range is performed.

[0553] Chroma scaling can correct the chroma signal based on the interrelationship between the luminance signal and the corresponding chroma signal.

[0554] Forward mapping can be performed between inter-prediction of the luminance signal and restoration of the luminance signal, and between inter-prediction of the luminance signal and chroma scaling. Inverse mapping can be performed between restoration of the luminance signal and in-loop filtering of the luminance signal. Chroma scaling can be performed between inverse transform and restoration of the chroma signal.

[0555] According to this structure, inverse quantizations for the luminance and chroma signals, inverse transforms for the luminance and chroma signals, prediction for the luminance signal, and restoration for the luminance signal can be performed within a mapped dynamic domain. In-loop filtering for the luminance and chroma signals, inter-predictions for the luminance and chroma signals, intra-predictions for the chroma signal, and restoration for the chroma signal can be performed within the existing dynamic domain.

[0556] A deblocking filter can remove block distortion occurring at the boundaries between blocks within the reconstructed image. For example, the blocks may be transformed blocks. Additionally, the blocks may be subblocks of a specific block described in the embodiments. Here, the boundaries between blocks may refer to samples adjacent to the boundaries between blocks.

[0557] A deblocking filter can be applied to the vertical and horizontal boundaries between blocks. After filtering is performed on the vertical boundaries of the blocks, filtering can be performed again on the horizontal boundaries of the filtered blocks.

[0558] A deblocking filter may be applied optionally. Whether to apply a deblocking filter to a target block may be determined based on at least one of sample(s) contained within a specific number of columns or rows within the target block and sample(s) contained within a specific number of columns or rows within a neighboring block adjacent to a specific boundary.

[0559] When a deblocking filter is applied to a target block, the filter to be applied may be determined according to the required strength of deblocking filtering. In other words, among a plurality of different filters, the filter determined according to the strength of deblocking filtering may be applied to the target block. The plurality of filters may include one of a long-tap filter, a strong filter, a weak filter, and a Gaussian filter.

[0560] The maximum length of the deblocking filter can be determined based on attributes of the target block, such as the size of the target block, the components of the target block, and coding parameters.

[0561] SAO can compensate for distortion between the original image and the reconstructed image on a sample basis. For compensation, SAO can apply an appropriate offset to the sample values. In other words, the offset can be added to the sample values.

[0562] An offset can be determined for the target block. For example, the offset can be determined for each component of the CTB. The determined offset can be applied to samples within a specific component of the CTB.

[0563] SAO may include an SAO using an edge offset (EO) and an SAO using a band offset (BO). Depending on the characteristics of samples within a specific block, such as a CTU, whether to perform an SAO using EO and whether to perform an SAO using BO may be determined, respectively.

[0564] In an SAO using EO, correction for distortion of samples can be performed based on the direction of edges within the target block. The pattern classes of the EO may include horizontal patterns, vertical patterns, 135-degree diagonal patterns, and 45-degree diagonal patterns. For the target block, information indicating the pattern class applied to the target block and multiple offsets of the said pattern class may be signaled. There may be four offsets. For a target sample within the target block, adjacent samples of the target sample may be determined according to the direction of the pattern class. An offset to be applied to the target sample may be determined by the pattern of the adjacent samples.

[0565] In an offset using BO, correction for sample distortion can be performed by classifying the brightness values ​​of samples within the target block into specific bands. The bit depth of the input image can be divided into m intervals. For example, m can be 32. The specific bands can be n consecutive intervals among the m intervals. For example, n can be 4. n offsets for the n intervals can be signaled. Additionally, information indicating the first interval selected as the n intervals among the m intervals can be signaled. The offset of the interval corresponding to the target sample can be added to the sample value of the target sample of the target unit.

[0566] ALF can compensate for distortion between the restored image and the original image.

[0567] The filter coefficients of the ALF can be signaled through the bitstream.

[0568] The filter shape of the ALF can be determined by the components of the target block. For example, a 7x7 diamond-shaped filter can be used for the luminance component. A 5x5 diamond-shaped filter can be used for the chroma component.

[0569] In ALF, the characteristics of a specific block can be determined for that block, and the class of that specific block can be determined based on those characteristics. In other words, the determination of characteristics and the determination of the class in ALF can be performed in units of 4x4 blocks. Filter coefficients can be calculated according to the class. A specific block can be a block with a size of 4x4.

[0570] One of 25 classes can be determined as the class of a specific block based on the direction and activity determined using the gradient of the specific block. Depending on the gradient of the specific block, a rotation transformation, a vertical reflection transformation, and / or a diagonal reflection transformation may be applied to the filter.

[0571] Information regarding whether ALF is applied can be signaled to specific units such as CTBs.

[0572] An index indicating a filter to be applied to a specific unit among the available filters may be signaled. Here, the available filters may include fixed filters and filters configured using a parameter set. For example, the parameter set may be an Adaptive Parameter Set (APS). The fixed filters may be predefined identically in the encoding device (110) and the decoder (150). The filter coefficients of the filters configured using the parameter set may be determined based on the coding parameters.

[0573]

[0574] Entropy Encoding and Entropy Decoding

[0575] Figure 7 shows entropy encoding and entropy decoding according to one example.

[0576] The processes of entropy encoding by the entropy encoder (139) are illustrated at the top of Fig. 7.

[0577] The entropy encoder (139) may include a context modeler, a binarization unit, and an entropy encoding unit. The context modeler may include a context selection unit and a context memory.

[0578] The binarization unit can generate binaries for syntactic elements by performing binarization on the syntactic elements of the target block. Binarization may be a process of converting syntactic elements into the form of binaries.

[0579] Information about syntax elements and beans can be provided from the binarization unit to the context selection unit.

[0580] The context modeler can perform context updates.

[0581] Context can refer to occurrence probability information for each bin regarding syntactic elements that have already been encoded.

[0582] The context modeler may perform a context update to apply current probability information to the entropy encoding of the bins of the syntactic elements of the target block. The updated context may be stored in context memory. At this time, the updated context corresponding to the syntactic elements of the target block (or the bins within the syntactic elements of the target block) may be derived by the context modeler.

[0583] The context selector can select a context corresponding to a bin of a syntactic element of a target block. The selected context can be loaded from context memory and used as an updated context for entropy encoding of the bins of the syntactic element of the target block.

[0584] The updated context can be used for entropy encoding of syntactic elements of the target block.

[0585] The entropy encoding unit can generate encoded information for syntactic elements of a target block by performing entropy encoding using generated bins and an updated context, and can generate a bitstream containing the encoded information. The entropy encoding unit may use at least one of an arithmetic encoding method and a bypass encoding method.

[0586] At the bottom of Fig. 7, the entropy decoding process by the entropy decoder (161) is illustrated.

[0587] The entropy decoder (161) may include a context modeler, an entropy decoder, and an inverse binary converter. The context modeler may include a context selection unit and a context memory.

[0588] The context modeler can perform context updates.

[0589] The context can refer to the probability information of occurrence for each bin regarding syntactic elements that have already been decoded.

[0590] The context modeler may perform a context update to apply the currently decoded probability information to the entropy decoding for the bins of the syntactic elements of the target block. The updated context may be stored in context memory. At this time, the updated context corresponding to the syntactic elements of the target block (or the bins within the syntactic elements of the target block) may be derived by the context modeler.

[0591] The context selection unit can select a context corresponding to a bin of a syntactic element of a target block. The selected context can be loaded from context memory and can be used as an updated context for entropy decoding of the syntactic element of the target block.

[0592] The updated context can be used for entropy decoding of the syntactic elements of the target block.

[0593] The entropy decoding unit can generate bins for the segmentation elements of the target block by performing entropy decoding on the encoded information of the bitstream based on the updated context. The entropy decoding unit may use at least one of an arithmetic decoding method and a bypass decoding method.

[0594] The debinaryization unit can obtain syntactic elements of a target block by performing debinaryization on at least one of the generated beans. Debinaryization may be a process of converting at least one of the beans into the form of a syntactic element.

[0595] Information about syntax elements and beans can be provided from the inverse binary unit to the context selector.

[0596] The syntax element may be one of the coding parameters described in the examples.

[0597]

[0598] Methods for binarization, inbinarization, entropy encoding, and entropy decoding

[0599] In the embodiments, to perform signaling for specific information, one or more of the binarization method, inverse binarization method, entropy encoding method and entropy decoding method listed below may be used.

[0600] - Signed 0-th order Exponential Golomb binarization / debinarization method (abbreviated as se(v))

[0601] - Signed k-order exponential-Golomb binarization / debinarization method (abbreviated as sek(v))

[0602] - 0-order exponentiation-Golomb binarization / debinarization method for unsigned positive integers (abbreviated as ue(v))

[0603] - k-order exponential-Golomb binarization / debinarization method for unsigned positive integers (abbreviated as uek(v))

[0604] - Fixed-length binarization / debinarization method (abbreviated as f(n))

[0605] - Truncated Rice binarization / debinarization method or truncated unary binarization / debinarization method (abbreviated as tu(v))

[0606] - Truncated binary binarization / debinarization method (abbreviated as tb(v))

[0607] - Context-adaptive arithmetic encoding / decoding method (abbreviated as ae(v))

[0608] - Bit string in bytes (abbreviated as b(8))

[0609] - Signed integer binary / debinary conversion method (abbreviated as i(n))

[0610] - Unsigned positive integer binarization / debinarization method (abbreviated as u(n)) ('u(n)' may also refer to a fixed-length binarization / debinarization method.)

[0611] - Unary Binary / Debinary Method

[0612]

[0613] Adaptive execution of the processing of the examples

[0614] The processing of the embodiments may be performed in the same and / or corresponding way in the encoding device (110) and the decoding device (150). Additionally, a combination of one or more of the above embodiments may be used for encoding and / or decoding of the image.

[0615] The order in which the embodiments are applied may differ from one another in the encoding device (110) and the decoding device (150). Alternatively, the order in which the above embodiments are applied may be the same (at least partially) in the encoding device (110) and the decoding device (150).

[0616] The processing of the embodiments may be performed for each of the specific targets. The processing of the embodiments may be performed identically for the specific targets. For example, the specific targets may include a luminance signal and a chroma signal.

[0617] The processes of the embodiments may be selectively applied / performed based on specific conditions or specific targets.

[0618] In one embodiment, the processing of the embodiment may be selectively applied / performed according to a temporal layer. Temporal layer information for a specific processing may be information indicating the temporal layer where the processing can be applied / performed. Temporal layer information may be signaled for a specific processing. Temporal layer information may indicate the lowest layer and / or the highest layer where the specific processing can be applied, and may indicate the specific layer where the specific processing is applied / performed. Alternatively, a fixed temporal layer where the processing of the embodiment is applied / performed may be defined.

[0619] In one embodiment, a type to which the processing of the embodiments is applied / performed may be defined, and whether the processing of the embodiments is applied / performed may be determined based on the defined type. The type may include a picture type, a slice type, and a tile group type, etc.

[0620] According to the description of the embodiments, when applying / performing a specific process on a specific target, specific conditions may be required, and the specific process may be performed under a specific decision. Where it is determined whether a specific condition is satisfied based on a specific coding parameter, or where a specific decision is made based on a specific coding parameter, such specific coding parameter may be interpreted as being replaceable with another coding parameter. That is to say, the coding parameter affecting the specific condition or specific decision described in the embodiments may be considered merely exemplary, and in addition to the specified coding parameter, one or more other coding parameters or a combination of one or more other coding parameters may be understood to perform the role of the specified coding parameter.

[0621] The processing of the embodiments may be applied / performed based on the size of at least one of the blocks described in the embodiments. For example, the blocks may include a coding block, a prediction block, a transformation block, a reference block, a current block, and a target block. Alternatively, the blocks may include adjacent blocks of the embodiments. Here, the size may be defined as a minimum size and / or a maximum size for the processing of the embodiments, or as a fixed size for the processing of the embodiments. Additionally, for the processing of the embodiments, a first embodiment may be applied at a first size, and a second embodiment may be applied at a second size. That is, the processing of the embodiments may be applied in combination depending on the size. Additionally, the processing of the embodiments may be applied only when the block size is greater than or equal to the minimum size and less than or equal to the maximum size. That is, the processing of the embodiments may be applied only when the block size falls within a specific range.

[0622]

[0623] The following describes improvements to in-loop filtering.

[0624] As described above, blocking artifacts and ringing artifacts present in a reconstructed image can be effectively reduced by performing at least one of deblocking filtering and SAO filtering on the reconstructed image generated by combining inter- or intra-predicted blocks and the reconstructed residual blocks. However, while the deblocking filter can reduce blocking artifacts occurring at the boundaries between blocks by performing vertical and horizontal filtering on the block boundaries, it has the disadvantage of not being able to minimize distortion between the original image and the reconstructed image during block boundary filtering. Additionally, although SAO partially minimizes distortion between the original image and the reconstructed image by utilizing rate-distortion optimization, it has limitations in terms of distortion minimization when the difference in distortion between the original image and the reconstructed image is large.

[0625] Neural network-based methods can be used to reduce distortion between the original image and the reconstructed image. These methods can be composed of neural networks based on convolution, transformers, deep residual blocks, serial / parallel connections, etc., that take multiple samples as input.

[0626] The information used as input to the neural network may be one or more of the reconstructed image, QP, slice QP, BS (boundary strength), Gradient, predictor sample, residual sample, HOG (histogram of oriented gradient), and slice mode (intra, inter), and may be any information available for use in the in-loop filtering encoding step without being limited thereto.

[0627] In-loop filtering according to the present disclosure may include at least one of neural net-based filtering, deblocking filtering, SAO, bilateral filtering, and ALF.

[0628] In-loop filtering can generate a decoded image by applying bidirectional filtering, deblocking filtering, neural network-based filtering, SAO, and ALF to the restored image in any order, and each filtering can be performed sequentially or two or more filterings can be performed in parallel in any order.

[0629] FIG. 8 is an exemplary diagram showing the structure of an in-loop filter according to one embodiment.

[0630] As shown in the example of Fig. 8, filtering can be performed on the restored image in the order of deblocking filtering, neural network-based filtering, SAO, and ALF.

[0631] However, the order in which in-loop filtering is performed is not limited to this.

[0632] For example, in-loop filtering can be applied to the reconstructed image in the order of neural network-based filtering, deblocking filtering, SAO, and ALF.

[0633] As another example, in-loop filtering can be applied to the reconstructed image in the order of bidirectional filtering, ALF, deblocking filtering, neural network-based filtering, and SAO.

[0634] As another example, in-loop filtering can be applied to the reconstructed image in the order of ALF, neural network-based filtering, deblocking filtering, and SAO.

[0635] For example, in-loop filtering can be applied to the reconstructed image in the order of ALF, SAO, deblocking filtering, and neural network-based filtering.

[0636] FIG. 9 is an exemplary diagram showing the structure of an in-loop filter according to another embodiment.

[0637] As shown in the example of Fig. 9, deblocking filtering and neural network-based filtering are executed in parallel, and then in-loop filtering can be applied to the reconstructed image in the order of SAO and ALF. The output of the deblocking filter and the output of the neural network-based filter can be combined through an average or weighted average and then input into the SAO.

[0638] Neural network-based loop filters are effective in reducing errors between the reconstructed image and the original image, and demonstrate high performance, particularly in reducing blocking artifacts. However, they have the disadvantage of high computational complexity and high memory usage during filtering.

[0639] The following disclosure relates to neural network-based filtering for overcoming the above disadvantages. In the following, unless otherwise noted, in-loop filtering or filtering may mean neural network-based filtering.

[0640] FIG. 10 shows a neural network-based filtering process performed by an image encoding device according to one embodiment, and FIG. 11 shows a neural network-based filtering process performed by an image decoding device according to one embodiment.

[0641] Referring to FIG. 10, the image encoding device can perform a filtering block classification process based on sample information (S1010), a filtering execution process (S1020), and a filter information encoding process (S1030).

[0642] Referring to FIG. 11, the image decoding device can perform a filter information decoding process (S1110), a filtering block classification process based on sample information (S1120), and a filtering execution process (S1130).

[0643] Here, the filtering block classification process (S1010, S1120) may be a process for determining whether to perform filtering on the filtering block.

[0644] Each process is described in detail below.

[0645]

[0646] 1. Filtering block classification based on sample information (S1010 & S1120)

[0647] Filtering can be performed on an NxM image region unit within the reconstructed image and / or difference image or other images. Hereinafter, the NxM image region is referred to as a 'filtering block'.

[0648] N and M can be positive integers. For example, N and M can be 256, 128, 64, 32, 16, 8, 4, or 2, respectively. N and M can have the same value or different values.

[0649] The filtering block may be at least one of CU, PU, ​​TU, CTU, slice, tile, frame (i.e., picture), and sequence.

[0650] Here, the frame type may be at least one of an I-frame, a P-frame, and a B-frame, and the filtering block size may be set for each frame type.

[0651] Here, the block size of the color difference block may differ from the size of the luminance block.

[0652] For example, the filtering block size for the luminance component can be N1xM1 and the filtering block size for the chrominance component can be N2xM2.

[0653] For example, the sizes of the color difference block and the luminance block can be changed by chroma subsampling.

[0654] It is possible to determine whether to perform filtering on filtering blocks through filtering block classification.

[0655] For example, an index C for filtering blocks can be derived through filtering block classification, and whether to perform filtering can be determined based on index C. K filters can be selected for filtering based on index C.

[0656] Here, C can be a positive integer.

[0657] Here, K represents the total number of filters and can be a positive integer.

[0658] K can be greater than the number of block indices C. Or, K can be equal to the number of block indices C. Or, K can be less than the number of block indices C.

[0659] For example, at least one of K filters can be selected in units of filtering blocks to perform filtering on the reconstructed image.

[0660] For example, U filters out of K filters can be selected in filtering block units to perform filtering on the reconstructed image.

[0661] For example, among the K filters in filtering blocks, U different filters can be selected for the luminance component filter and the chrominance component filter, respectively, to perform filtering on the reconstructed image.

[0662] In one embodiment, a filtering block index C can be derived through filtering block classification, and whether to perform filtering can be determined using this. For example, if the filtering block is a 256x256 block, filtering may or may not be performed in units of 256x256 blocks according to index C. Alternatively, if the filtering block is a CTU, filtering may or may not be performed in units of CTUs according to index C.

[0663] In one embodiment, a filtering block index C can be derived through filtering block classification, and the index can be used to determine whether to perform luminance component filtering and / or chrominance component filtering.

[0664] For example, based on index C, filtering for the luminance block may be performed on a filtering block basis, and filtering for the chrominance block may not be performed.

[0665] As another example, based on index C, filtering for the color difference block may be performed on a filtering block basis, and filtering for the luminance block may not be performed.

[0666] Block classification method

[0667] Classification of filtering blocks can be derived based on sample information. In other words, classification parameters for classifying filtering blocks can be derived based on sample information.

[0668] Here, the sample information may be information representing sample values ​​within the restored image. Alternatively, it may be information used for encoding / decoding the samples (e.g., prediction information, transformation information, segmentation information, etc.).

[0669] In some embodiments, classification of filtering blocks may be performed based on a boundary strength (BS) determined by at least one of prediction information or transformation information of two blocks forming a boundary to perform deblocking filtering.

[0670] BS can be derived based on the encoding information of adjacent P and Q blocks as shown in Fig. 12.

[0671] BS can have 0 or a positive integer. For example, BS can have values ​​of 0, 1, or 2.

[0672] P and Q blocks can be CTU, CU, TU, or PU units.

[0673] Blocks P and Q can be blocks of size NxM. Here, N and M can be positive integers (=1, 2, 3, 4,...).

[0674] For example, as in (A) of FIG. 12, P and Q may be 8x4 blocks, or as in (B) of FIG. 12, P and Q may be 8x8 blocks.

[0675] Here, blocks P and Q can be in left-right or up-down positions.

[0676] BS can be set to 2 depending on the condition between blocks P and Q.

[0677] For example, if the P or Q block is in intra mode, BS can be set to 2.

[0678] For example, if the P or Q block is in Combined Intra / Inter-Picture Prediction (CIIP) mode, BS can be set to 2.

[0679] BS can be set to 1 depending on the condition between blocks P and Q.

[0680] For example, if there is a difference in motion vectors between blocks P and Q, BS can be set to 1.

[0681] For example, if there is a non-zero transformation factor in block P or Q, BS can be set to 1.

[0682] For example, BS can be set to 1 if the prediction modes of blocks P and Q are different. Here, the prediction mode can be inter, palette, or intra-picture block copy (IBC).

[0683] For example, if the prediction mode of blocks P and Q is IBC and the IBC vectors are different, BS can be set to 1.

[0684] BS can be set to 0 except for the above conditions between P and Q blocks. If BS is 0, deblocking filtering may not be performed.

[0685] With the BS obtained from the above conditions, a BS boundary line frame or a BS area frame of the same size as the restored frame can be created.

[0686] For example, C which can have different values ​​for the BS value i can be multiplied. Here, C i can be a real number, and i can be an integer. For example, when BS is 1, C1 can be multiplied, and when BS is 2, C2 can be multiplied. Here, C1 and C2 can be the same or different values.

[0687] As another example, C, which can have different values ​​for the BS value. i can be added. Here, C i can be a real number. For example, when BS is 1, C1 can be added, and when BS is 2, C2 can be added. Here, C1 and C2 can be the same or different values.

[0688] Block Q may be at least one of blocks Q3, Q2, Q1, and Q0 adjacent to Block P, as exemplified in FIG. 13.

[0689] Blocks P and Q of Fig. 13 are P 0a , P 0b , P 0c , P 0d , D 0a , D 0b , D 0c , D 0d , Q 3a , Q 2b , Q 1c , Q 0d A frame can be created in which the obtained BS value is stored in the area that is accessible in at least one of the areas, and 0 is stored in the P and Q blocks that do not correspond to this.

[0690] Here, D 0a , D 0b , D 0c , D 0dIn at least one of the regions, P 0a , P 0b , P 0c , P 0d Values ​​derived from BS values ​​stored in at least one of the regions may be stored. For example, the maximum, minimum, or average value of the BS values ​​may be stored.

[0691] As another example, P block and Q block are P 0a , P 0b , D 0a , D 0b , D 0c It can be accessed in at least one of the regions, and the calculated BS value is stored in the accessed region, and the remaining Q 1c , Q 0d You can create a frame that stores 0 in at least one of the locations.

[0692] Here, D 0a , D 0b , D 0c In at least one of the regions, P 0a , P 0b The maximum, minimum, or average value of the BS value stored in at least one of the areas may be stored.

[0693] As another example, P block and Q block are P 0c , P 0d , D 0b , D 0c , D 0d It can be accessed in at least one of the regions, and the calculated BS value is stored in the accessed region, and the remaining Q 3a , Q 2b You can create a frame that stores 0 in at least one of the locations.

[0694] Here, D 0b , D 0c , D 0d In at least one of the regions, P 0c , P0d The maximum, minimum, or average value of the BS values ​​stored in at least one of the regions may be stored.

[0695] As another example, the maximum value among the BS values ​​obtained from at least one of the Q blocks (Q3, Q2, Q1, Q0) can be stored as a value in the entire P block.

[0696] Filtering blocks can be classified using BS boundaries.

[0697] The BS boundary line can be in pixel units and / or in pixel blocks of size HxW. Here, H and W can be positive integers.

[0698] The BS boundary line can be composed of samples having a BS value of 1 or greater.

[0699] For example, the BS boundary can be composed of samples having a BS value of 1 or 2. That is, the BS boundary can be composed of samples having a BS value that is not 0.

[0700] As another example, the BS boundary line can be composed of samples having a BS value of 1.

[0701] As another example, the BS boundary line can be composed of samples having a BS value of 2.

[0702] As an example of classification based on BS boundaries, the number of BS partitions within a filtering block may be used.

[0703] A BS partition refers to the area enclosed by the BS boundary line when the BS boundary line of a frame is drawn.

[0704] FIG. 14 is an illustrative diagram for explaining a BS partition according to one embodiment.

[0705] In Fig. 14, the shaded area represents the BS boundary line within the filtering block.

[0706] Figure 14 (A) is an example in which a filtering block is divided into 4 BS partitions by BS boundary lines, and (B) is an example in which a filtering block is divided into 3 BS partitions.

[0707] As another embodiment of classification based on BS boundaries, the number of BS samples may be used. The number of BS boundary samples may refer to the number of samples constituting the BS boundaries.

[0708] FIG. 15 is an example diagram illustrating the number of BS samples according to one embodiment.

[0709] In Fig. 15, the shaded area represents the BS boundary line within the filtering block.

[0710] Figure 15 (A) illustrates a case where the number of BS samples in the filtering block is 54, and (B) illustrates a case where the number of BS samples in the filtering block is 64.

[0711] For example, the number of BS samples can be the number of samples having a BS value of 1 or 2. That is, the number of BS samples can be the number of samples having a BS value that is not 0.

[0712] As another example, the number of BS samples may be the number of samples that have a BS value of 1.

[0713] As another example, the number of BS samples may be the number of samples having a BS value of 2.

[0714] As another embodiment of classification based on BS boundaries, it can be used for statistical values ​​of BS values. Here, the statistical value may be, for example, the sum of the BS values.

[0715] FIG. 16 is an example diagram illustrating the total sum of BS values ​​according to one embodiment.

[0716] In the filtering block shown in Fig. 16, the white parts are parts where the BS value is 0, the hatched parts are boundary lines where the BS value is 1, and the parts filled with dots can be boundary lines where the BS value is 1.

[0717] For example, the sum of the BS values ​​may be the sum of the BS values ​​on the boundary line having a BS value of 1 or 2. That is, it may be the sum of non-zero BS values. For example, referring to FIG. 16, the sum of the BS values ​​may be 96.

[0718] As another example, the sum of the BS values ​​may be the sum of the BS values ​​on the boundary line having a BS value of 2. For example, referring to FIG. 16, the sum of the BS values ​​may be 64.

[0719] Alternatively, the sum of the BS values ​​may be the sum of the BS values ​​on the boundary line having a BS value of 1. For example, referring to FIG. 16, the sum of the BS values ​​may be 32.

[0720] Classification of filtering blocks can be performed using at least one of the number of BS partitions, the number of BS samples, or the sum of the BS values ​​as a classification parameter.

[0721] For example, one or more of the aforementioned classification parameters are given threshold values ​​T1, T1, ..., T corresponding to the classification parameters. i A classification index C for each filtering block can be derived by comparison with. Here, i can be a natural number greater than or equal to 1.

[0722] threshold T j can be any positive integer. T j can be 1, 2, 4, 8 or other integer values. In this case, j is 1, 2, ..., i.

[0723] The magnitudes of the threshold values ​​Tj are T1 ≤ T2 ≤ ... ≤ T i It can be set as follows.

[0724] The number of BS partitions, the number of BS samples, or the total BS of the filtering block is T j-1 Larger or T jClassification index C corresponding to the case where it is smaller can be derived.

[0725] In some embodiments, classification of a filtering block may be performed based on at least one of the number of CUs, the number of TUs, or the number of PUs within the filtering block. That is, at least one of the number of CUs, the number of TUs, or the number of PUs within the filtering block may be used as a classification parameter.

[0726] For example, at least one classification parameter among the number of CUs, the number of TUs, or the number of PUs is a threshold T1, T1, ..., T i A classification index C can be assigned to each filtering block by comparison with. For example, the number of CUs, TUs, or PUs of the filtering block is T j-1 Larger or T j Classification index C corresponding to the case where it is smaller can be derived.

[0727] In some embodiments, classification of filtering blocks may be performed based on sample edges.

[0728] As an example of filtering block classification based on sample edges, the edge size (amplitude) e ij It can be used.

[0729] FIG. 17 is an illustrative diagram for explaining sample edge size calculation according to one embodiment.

[0730] The edge size is 2 samples S i and S j It can be the difference between or the absolute value of that difference. That is, e ij = S i - S j or e ij = |S i - S j It can be.

[0731] S i and S jIt can be a luminance component sample and a color difference component sample.

[0732] As illustrated in (A) to (C) of FIG. 17, S used in edge size calculation i and S j can be samples adjacent to each other in at least one of the up, down, left, right, or diagonal positions.

[0733] Or, as illustrated in (D) to (F) of FIG. 17, S used in edge size calculation i and S j may be samples that are not adjacent to at least one of the up, down, left, right, or diagonal positions.

[0734] Depending on the orientation in which the samples are placed, the edge size in that direction can be calculated.

[0735] Figures 17 (A) and (D) show examples of samples used for calculating vertical edge sizes, (B) and (E) show examples of samples used for calculating horizontal edge sizes, and (C) and (F) show examples of samples used for calculating diagonal edge sizes. Figure 17 illustrates only downward diagonals (Figures 17 (C) and (F)), but upward diagonals may be used in place of downward diagonals, or upward diagonals may be used additionally.

[0736] Sample S used in edge size calculation i can be all samples and / or some samples of the filtering block.

[0737] FIG. 18 is an example of samples used for calculating sample edge sizes according to one embodiment. In FIG. 18, "x" indicates the sample location where the edge size is calculated.

[0738] As in the example of (A) in Fig. 18, sample S used for edge size calculation i can be any sample within the filtering block.

[0739] As shown in examples (B) to (G) of FIG. 18, sample S used for edge size calculation i can be some samples of the filtering block.

[0740] For filtering block classification, the sum or average of sample edge sizes can be used. That is, filtering blocks can be classified using the sum or average of edge sizes as a classification parameter.

[0741] At least one of the vertical, horizontal, and diagonal directions can be selected for the edge direction. The edge size sum or average can be calculated for at least one of the vertical, horizontal, and diagonal directions. Alternatively, after calculating the edge size sums (or averages) for each direction, the maximum, minimum, or average of the edge size sums corresponding to each direction can be selected as the final edge size sum for filtering block classification.

[0742] The edge size sum E is the edge size e calculated for the filtering block. ij It can be a union of.

[0743] Edge size sum E and threshold T1, T1, ..., T for edge size sum i A classification index C for the filtering block can be assigned by comparing them. For example, if E is T j-1 Larger or T j Classification index C corresponding to the case where it is smaller can be derived.

[0744] As another embodiment of filtering block classification based on sample edges, edge directionality can be utilized. That is, filtering blocks can be classified using edge directionality as a classification parameter.

[0745] In deriving the edge orientation of a KxJ block, at least one of the vertical and / or horizontal and / or diagonal edges of samples within the block may be considered.

[0746] For example, the sum and / or sum and / or average of the edge sizes corresponding to at least one direction among the vertical, horizontal, and diagonal edges of the samples inside the block can be calculated, and at least one of the edge directions having the largest value can be used as the edge directionality of the block.

[0747] Here, the block may be a filtering block or an internal sub-block of the filtering block.

[0748] Here, K and J can be positive integers (=1, 2, ...).

[0749] FIG. 19 is an illustrative diagram for explaining the derivation of edge directionality according to one embodiment. In FIG. 19, "x" indicates a sample location where the edge size is calculated.

[0750] Figure 19 (A) is an example diagram showing edge directionality in 2x2 sub-block units within a filtering block.

[0751] As illustrated in (A) of FIG. 19, the sum of vertical edge sizes E in 2x2 sub-block units within the filtering block v , sum of horizontal edge sizes E h , sum of diagonal edge sizes E d It is possible to obtain the edge directionality of a 2x2 sub-block based on the sum of the edge sizes for each direction.

[0752] For example, E v , E h , E d The direction corresponding to the largest value among them may be the edge directionality of the 2x2 sub-block. Or, E v , E h , E d The directions corresponding to the first and second largest values ​​among them may be the edge directions of the 2x2 sub-block.

[0753] Figures 19 (B) to (D) are example diagrams showing edge directionality in 4x4 sub-block units within a filtering block.

[0754] As shown in (B) of FIG. 19, the edge size can be calculated at all sample locations within a 4x4 sub-block. Alternatively, as shown in (C) or (D) of FIG. 19, the edge size can be calculated at some sample locations within a 4x4 sub-block.

[0755] E calculated for a 4x4 sub-block v , E h , E d The direction corresponding to the largest value among them may be the edge directionality of the 4x4 sub-block. Or, E v , E h , E d The directions corresponding to the first and second largest values ​​among them may be the edge directions of the 4x4 sub-block.

[0756] Based on the edge directionality of each sub-block, the number of sub-blocks Di having an edge directionality in direction i can be derived as the directionality parameter in direction i. Filtering blocks can be classified based on Di.

[0757] i can have integer values, and for example, each integer value can correspond to a vertical, horizontal, or diagonal direction.

[0758] For example, the number of subblocks with the highest frequency of directionality (i.e., the maximum value among Di) is the final directionality parameter D of the filtering block. final It can be derived as. The classification of filtering blocks is D final It can be performed using as a classification parameter. For example, D final This threshold T j-1 Larger or T j Classification index C corresponding to the case where it is smaller can be derived.

[0759] As another example, D finalIt can be derived through the weighted sum of Di for each direction. For example, a*D h + b*D v + C*D d D by + d final This can be derived. Here, D h , D v , and D d represents the number of sub-blocks corresponding to the horizontal, vertical, and diagonal directions, respectively.

[0760] Here, equal weights can be used for the weighted sum.

[0761] Alternatively, weights for the weighted sum may be derived based on the sum or average of the edge size sums of the sub-blocks corresponding to each direction. For example, for sub-blocks having the same direction, the sum or average of the edge size sums for that direction may be calculated. The weights for each direction may be set in proportion to the sum or average of the edge size sums for that direction.

[0762] In some embodiments, classification of filtering blocks may be performed based on neural network-based filtering information for reference blocks. For example, an activation map may be generated using neural network-based filtering information for each sample (i.e., pixel) within reference blocks referenced by motion information of inter-predicted blocks. Block classification may be performed based on the activation map. After reducing the activation map, it may be determined whether to perform filtering on the corresponding block through comparison with a threshold value.

[0763] For example, neural network-based filtering information for each sample within the reference blocks may refer to a mode index representing the neural network-based filtering mode applied to each sample within the reference blocks. The neural network-based filtering mode may be selected from among a plurality of modes, and an index corresponding to each mode may be assigned. This will be described later with reference to FIG. 24.

[0764] To generate an activation map, the BS values ​​of each sample within the filtering block may also be used. For example, the values ​​for each sample in the activation map can be derived through operations between the mode index and the BS value for each sample (e.g., sum, average, or weighted average).

[0765] The reduced activation map can be the sum or average of the values ​​assigned to each sample of the activation map.

[0766] As described above, block classification can be performed by using at least one of the following as classification parameters: the number of BS partitions, the total sum of BS, the number of BS boundary samples, the number of at least one of CU, PU, ​​or TU within the filtering block, the sum of sample edge sizes, or edge directionality, or reference block information.

[0767] A filtering block can be classified based on at least one classification parameter. For example, a classification index C can be derived through a comparison between a classification parameter derived for a filtering block and at least one threshold value corresponding to that classification parameter. Depending on the index C, neural network-based filtering may or may not be performed on the filtering block.

[0768] For example, the classification parameter for the filtering block is a specific threshold T j If it is greater than or equal to, filtering for the corresponding filtering block may not be performed. In other words, if the classification parameter is a specific threshold T j Filtering may not be performed on filtering blocks having a classification index C that is greater than or equal to.

[0769] For example, the classification parameter for the filtering block is a specific threshold T j If it is less than or equal to, filtering for the corresponding filtering block may not be performed. In other words, if the classification parameter is a specific threshold Tj Filtering may not be performed on filtering blocks having a classification index C that is less than or equal to.

[0770] For example, the classification parameter for the filtering block is a specific threshold T j-1 Larger than T j If it is less than or equal to, filtering for the corresponding filtering block may not be performed. In other words, if the classification parameter is a specific threshold T j-1 Larger than T j Filtering may not be performed on filtering blocks having a classification index C that is less than or equal to the case.

[0771] Threshold for block classification

[0772] Threshold values ​​for classification parameters (or threshold values ​​used to derive classification indices) T1, T2, ..., T i It can be produced adaptively.

[0773] The threshold is D or N F It can be derived based on at least one of these. This method is defined as Time-Distortion Optimization (TDO).

[0774] For example, it can be derived based on the Lagrange multiplier method, such as the following formula.

[0775] Cost = D + λ*N F

[0776] Here, D can be the error between the original block and the restored block. Or, D can be the error between the original block and the restored block to which the transformation was applied. For example, it can be SATD (sum of absolute transformed differences).

[0777] The restoration block can be a filtered restoration block or a restoration block with no filtering applied.

[0778] Here, N F This may be the time during which neural network-based filtering is performed.

[0779] Or, N F This could be the time it takes to encode a frame.

[0780] Or, N F may be the number of blocks on which neural network-based filtering is performed.

[0781] Here, the threshold T used for filtering block classification. j If you change the value, N F The value can change. N, which yields the minimum cost. F T corresponding to j The optimal number of filtering blocks can be derived using this. In this case, j can be a positive integer.

[0782] Here, λ can be any real value. For example, it can be 10000, 20000, or 30000. By adjusting λ, the number_of_enabled_block with the minimum Cost can be found.

[0783] λ can be determined adaptively.

[0784] For example, λ can have different values ​​depending on the base QP (e.g., sequence or picture level QP) or the slice level slice QP.

[0785] For example, the range of QP can be divided into multiple intervals by multiple QP thresholds. Different λ values ​​may be used depending on which interval the slice QP belongs to. The larger the slice QP, the larger the λ value may be.

[0786] For example, multiple QP thresholds can be QP1, QP2, QP3, QP4, QP5 (QP1 < QP2 < QP3 < QP4 < QP5). For instance, QP1=24, QP2=27, QP3=32, QP4=37, QP5=42.

[0787] Here, different λ values ​​may be used depending on which of the six intervals distinguished by QP1, QP2, QP3, QP4, and QP5 it belongs to. For example, the λ value can be set as follows.

[0788] - If slcie QP < QP1, then λ = λ1

[0789] - If QP1 ≤ slice QP < QP2, λ= λ2(λ2> λ1)

[0790] - If QP2 ≤ slice QP < QP3, λ= λ3(λ3> λ2)

[0791] - If QP3 ≤ slice QP < QP4, λ= λ4(λ4> λ3)

[0792] - If QP4 ≤ slice QP < QP5, λ= λ5(λ5> λ4)

[0793] - If QP5 ≤ slice QP, then λ= λ6(λ6> λ5)

[0794] In this example, it was described that a total of five threshold values ​​are used, but the present invention is not limited thereto. For example, by additionally setting at least one QP threshold value greater than QP5, different λ values ​​can be used depending on which section the slice QP belongs to, even within a large QP range.

[0795] λ can have different values ​​for each filtering block size.

[0796] λ can have different values ​​for each filtering frame size.

[0797] λ can have different values ​​for each temporal layer.

[0798] λ can have different values ​​for each input video.

[0799] λ can have different values ​​in GOP (Group of Pictures) units.

[0800] The video encoding device can determine a threshold value and signal information about the threshold value to the video decoding device.

[0801] The determination of the threshold can be performed at any stage of neural network-based in-loop filtering.

[0802] FIG. 20 is an example diagram showing the location where a threshold value is determined among the neural network-based in-loop filtering steps according to one embodiment.

[0803] Neural network-based in-loop filtering is performed in the order of Filtering, Parameter Search, and RDO (Rate-Distortion Optimization).

[0804] The determination of the threshold value can be performed in the middle of the neural network-based in-loop filtering process, as shown in (A) of FIG. 20. Or, it can be performed at the end of the neural network-based in-loop filtering process, as shown in (B) of FIG. 20. Or, it can be performed before the neural network-based in-loop filtering process, as shown in (C) of FIG. 20.

[0805] Threshold values ​​T1, T2, ..., T obtained for classifying the filtering blocks above i They can be composed of a lookup table set.

[0806] For example, i threshold values ​​T j They can be classified into M (=1, 2, ...) lookup tables, each having N (=1, 2, ...) threshold values ​​as elements.

[0807] For example, four thresholds T1, T2, T3, T4 can be configured into two lookup tables, lookup table #1 {T1, T2} and lookup table #2 {T3, T4}. Here, #1 and #2 may be unique numbers of the lookup tables.

[0808] Each lookup table may have a unique number X. The unique number X can be derived by a filtering classification method, i.e., a classification parameter.

[0809] For example, at least one of the number of BS (boundary strength) partitions, the number of BS samples, and the total sum of BS values ​​may be used, and a unique number may be assigned to each method.

[0810] For example, at least one of the edge size sum and edge directionality may be used, and a unique number may be assigned to each method.

[0811] For example, at least one of the comparison methods for the number of CUs, the number of TUs, or the number of PUs within the filtering block may be used, and a unique number may be assigned to each method.

[0812] The unique number X of the lookup table can be derived from at least one of a slice type, QP, slice QP, temporal ID, and gradient value of a filtering block, or any information available in the in-loop filtering encoding and / or decoding step, without being limited thereto.

[0813] For example, a fixed number of a lookup table can be assigned to at least one unit among I-slice, B-slice, and P-slice.

[0814] For example, at least one lookup table fixed number can be mapped to a temporal hierarchy identifier.

[0815]

[0816] 2. Perform filtering (S1020 & S1130)

[0817] Filtering can be performed on the restored image corresponding to the block determined to be filtered based on the filtering block classification.

[0818] When filtering is performed, K filters corresponding to the block index C determined for the filtering block may be selected.

[0819] For example, at least one of K filters is selected in a filtering block, and filtering can be performed on a filtering block basis.

[0820] For example, U filters out of K filters can be selected for the filtering block corresponding to filtering block index C, and filtering can be performed on a filtering block basis. Here, U can be a positive integer and can be 1, 2, ..., K.

[0821] The K filters may be a filter set including neural network-based filters and / or multiple filters.

[0822] The K filters can form the structure of a filtering network in a stacked form with at least one layer. For example, at least one of a convolution layer, an activation layer, a pooling layer, a normalization layer, and a dropout layer may be used.

[0823] The number of layers can be L. In this case, L can be a positive integer.

[0824] K filters can form at least one filtering network for each of the brightness component (luma) and the color difference component (chroma), and can be connected in series and / or in parallel.

[0825] K filters may differ from each other in at least one of the number of layers, layer types, layer order, and learning method.

[0826] Whether filtering based on filtering block classification is allowed or enabled can be determined at a level of a region higher than the filtering block. The level of the higher region may be at least one of CU, CTU, slice, tile, frame, or sequence.

[0827] For example, if it is determined that filtering based on filtering block classification is not allowed in a higher-level area, filtering based on filtering block classification is not performed for filtering blocks belonging to that higher-level area.

[0828] Meanwhile, if it is determined that filtering based on filtering block classification is allowed in a higher-level area, the decision on whether to perform filtering for filtering blocks belonging to the higher-level area can be made based on the filtering block classification. For example, assuming the higher-level area is a slice and the filtering block is a CTU, even if it is determined at the slice level that filtering is allowed for all CTUs within the slice, filtering may not be performed for at least one CTU within the slice.

[0829] When the number of filtering blocks that are not filtered by filtering block classification within the upper region is greater than or equal to the threshold D, filtering may not be performed on all filtering blocks within the upper region. Here, D can be 0 or a positive integer.

[0830] On the other hand, if the number of filtering blocks for which filtering is not performed is smaller than the threshold D, the filtering status of each filtering block within the upper region can be determined based on the block classification result for that filtering block.

[0831] FIGS. 21 to 23 are exemplary diagrams for explaining whether to perform filtering on filtering blocks included in an upper region according to one embodiment.

[0832] In FIGS. 21 to 23, (A) is an example diagram showing the classification results for filtering blocks within the upper region, and the hatched area represents a filtering block determined not to have filtering performed. In FIGS. 21 to 23, (B) represents the final decision on whether to perform filtering on the filtering blocks within the upper region.

[0833] For example, if the filtering block is in CU units, when there are D or more CUs in a CTU where filtering is not performed, filtering may not be performed on all CUs in that CTU.

[0834] As shown in FIG. 21, when the threshold D is 7 and there are 8 CUs for which filtering is not performed, filtering may not be performed on all CUs of the corresponding CTU.

[0835] For example, if the filtering block is in CTU units, when there are D or more CTUs in a slice for which filtering is not performed, filtering may not be performed on all CTUs in that slice.

[0836] As shown in FIG. 22, when the threshold D is 6 and there are 9 CTUs for which filtering is not performed, filtering may not be performed on all CTUs in the corresponding slice.

[0837] For example, if the filtering block is in slice units, when there are D or more slices in a frame for which filtering is not performed, filtering may not be performed on all slices in that frame.

[0838] As shown in FIG. 23, when the threshold D is 1 and there are 2 slices where filtering is not performed, filtering may not be performed on all slices of the frame.

[0839]

[0840] 3. Encoding / decoding step of filter information (S1030 & S1110)

[0841] At least some of the filter information can be signaled at the filtering block level or at a level higher than the filtering block.

[0842] For example, at least a portion of the filter information can be entropy encoded and / or decoded at the level of the first CTU within a slice in the bitstream.

[0843] For example, at least a portion of the filter information can be entropy encoded and / or decoded in at least one unit among the sequence parameter set, picture parameter set, slice header, and slice data within the bitstream.

[0844] For entropy encoding and / or decoding of at least one of the filter information, at least one of the binarization methods described above may be used.

[0845] The above filter information may include at least one of the following: information on whether to perform luminance component filtering, information on whether to perform chrominance component filtering, neural network-based filter type information, filtering block type information, filtering parameter information, information on whether to perform filtering of at least one unit among CU / CTU / slice / tile / picture / sequence, neural network-based filter information, information on whether to use different filters for luminance components and chrominance components, information on whether to perform filtering block classification, filtering block classification index information, BS partition threshold information, BS count threshold information, and BS total sum threshold information.

[0846] Information on whether filtering is performed can be signaled / encoded / decoded at least one level among slice, block, and frame.

[0847] The neural network-based filter shape information may be the filter information of at least one of K neural network-based filters. Here, K may be a positive integer.

[0848] Filtering block shape information may include at least one of the width and height of the filtering block.

[0849] Alternatively, filtering block-shaped information can be derived without additional information during the encoding and / or decoding stages.

[0850] For example, if the number of CTUs N of a slice is greater than or less than a specific value K, the filtering block size H x W can be derived. Here, N, K, H, and W can be integers.

[0851] For example, if the number of frame slices N is greater than or less than a specific value K, the filtering block size H x W can be derived. Here, N, K, H, and W can be integers.

[0852] Filtering parameter information may include information on whether a filter is applied at least one level among a filtering block, slice, frame, or sequence.

[0853] The filtering parameter information refers to the weight values ​​w = {w0, w1, ... , w} required for the reconstructed samples and / or the output samples of the deblocking filter and / or the prediction samples and / or the difference samples and / or the output samples of the neural network-based filter to be weighted summed. i It can include}.

[0854] Here, the weight w can have at least one value and can be a real number. i can be a positive integer.

[0855] The weight w can be determined at least one level among CU, PU, ​​TU, CTU, filtering block, slice, frame, and sequence.

[0856] The weight w can use at least one value for each filtering block.

[0857] For example, at least one weight w can be used for each CU and / or TU and / or PU unit.

[0858] For example, a different weight w can be used for at least one unit among CTU, filtering block, slice, frame, and sequence.

[0859] In some embodiments, the weight w may be a fixed value.

[0860] For example, a unique number can be assigned to each fixed value so that only the unique number can be encoded and / or decoded. Here, at least one available fixed value may be stored in the encoder and / or decoder.

[0861] As another example, fixed weights w can be derived without encoding and / or decoding. Weights w can be derived based on classification parameters.

[0862] For example, if the number of BS partitions in the filtering block is N or more, a predetermined fixed weight value W can be used. Here, N can be an integer and W can be a real number.

[0863] For example, if the number of BS samples in the filtering block is N or more, a predetermined fixed weight value W can be used. Here, N can be an integer and W can be a real number.

[0864] For example, if the sum of the BS samples of a filtering block is greater than or equal to N, a predetermined fixed weight value W can be used. Here, N can be an integer and W can be a real number.

[0865] For example, if there are M or more filtering blocks in a slice where the number of BS partitions is N or more, a predetermined fixed weight value W can be used. Here, N and M can be integers, and W can be a real number.

[0866] For example, if there are M or more filtering blocks in a slice where the number of BS samples is N or more, a predetermined fixed weight value W can be used. Here, N and M can be integers, and W can be a real number.

[0867] For example, if there are M or more filtering blocks in a slice where the BS sample sum is N or greater, a predetermined fixed weight value W can be used. Here, N and M can be integers, and W can be a real number.

[0868] For example, if there are S or more slices in a frame where there are M or more filtering blocks with N or more BS partitions, a predetermined fixed weight value W can be used. Here, N, M, and S can be integers, and W can be a real number.

[0869] For example, if there are S or more slices in a frame, each having at least M filtering blocks with N or more BS samples, a predetermined fixed weight value W can be used. Here, N, M, and S can be integers, and W can be a real number.

[0870] For example, if there are S or more slices in a frame where there are M or more filtering blocks with a BS sample sum of N or more, a predetermined fixed weight value W can be used. Here, N, M, and S can be integers, and W can be a real number.

[0871] For example, if the sum of the edge sizes of a filtering block is greater than or equal to E, a predetermined fixed weight value W can be used. Here, E can be an integer and W can be a real number.

[0872] For example, if the average edge size of a filtering block is greater than or equal to E, a predetermined fixed weight value W can be used. Here, E can be an integer and W can be a real number.

[0873] For example, if the edge direction parameter in the filtering block is greater than or equal to D, a predetermined fixed weight value W can be used. Here, E can be an integer and W can be a real number.

[0874] For example, if the sum of edge sizes in a slice is greater than or equal to E, a predetermined fixed weight value W can be used. Here, E can be an integer and W can be a real number.

[0875] For example, if the average edge size in a slice is greater than or equal to E, a predetermined fixed weight value W can be used. Here, E can be an integer and W can be a real number.

[0876] For example, if the maximum value among the edge direction parameters of each direction in a slice is greater than or equal to D, a predetermined fixed weight value W can be used. Here, E can be an integer and W can be a real number.

[0877] For example, if the sum of edge sizes in a frame is greater than or equal to E, a predetermined fixed weight value W can be used. Here, E can be an integer and W can be a real number.

[0878] For example, if the average edge size in a frame is greater than or equal to E, a predetermined fixed weight value W can be used. Here, E can be an integer and W can be a real number.

[0879] For example, if the maximum value of the edge directionality parameters for each direction in a frame is greater than or equal to D, a predetermined fixed weight value W can be used. Here, E can be an integer and W can be a real number.

[0880] As another example, a fixed weight w can be derived based on a feature map inside a neural network-based filter.

[0881] For example, if the total sum of feature map samples within a neural network-based filter is greater than or less than M, a predetermined fixed weight value W can be used. Here, M and W can be real numbers.

[0882] For example, if the number of samples having values ​​between N and M in a feature map within a neural network-based filter is greater than or less than K, a predetermined fixed weight value W can be used. Here, N and M can be real numbers, and K can be an integer.

[0883] At least one unit difference value R among CU, PU, ​​TU, CTU, filtering block, slice, and frame can be added to the fixed weight value w of the filtering parameter above. The final weight used in the weighted sum may be w + R. Here, the difference value R may be a real number.

[0884] The difference value R can be encoded and / or decoded in an encoder and / or decoder. Alternatively, the difference value R can be derived without additional information.

[0885] The difference value R may be the same or different in at least one unit among CU, PU, ​​TU, CTU, filtering block, slice, and frame.

[0886] For example, when weight values ​​w are applied at the slice level, different difference values ​​R0, R1, ..., R are applied for each filtering block within the slice. i It can be applied.

[0887] For example, when applying weight values ​​w on a slice basis, the same difference value R can be applied to each filtering block within the slice.

[0888] For example, when weight values ​​w are applied on a frame-by-frame basis, different difference values ​​R0, R1, ..., R exist for each slice within the frame. i It can be applied.

[0889] For example, when applying weight values ​​w on a frame-by-frame basis, the same difference value R can be applied to each slice within the frame.

[0890] In some embodiments, the weight w may be an adaptive weight value.

[0891] Adaptive weight values ​​can be encoded and / or decoded.

[0892] Alternatively, adaptive weight values ​​can be derived without encoding or decoding.

[0893] Adaptive weight values ​​can be derived based on at least one of the aforementioned classification parameters.

[0894] As an exemplary embodiment, the adaptive weight value may be calculated using at least one of the number of BS partitions, the number of BS samples, and the total sum of BS of the filtering block.

[0895] Below, the number of BS partitions in the filtering block is N p , BS number of samples N s , the sum of BS is N a It is displayed. Assume that N is the value obtained from one of the methods among the number of BS partitions, the number of BS samples, and the total sum of BS in the filtering block.

[0896] For example, the ratio and / or sum and / or difference and / or product between N and a specific value K can be used as an adaptive weight value.

[0897] As another example, N p , N s , and N a Adaptive weight values ​​can be derived using the ratio of two of the values ​​and a specific value K.

[0898] For example, N p and N s The ratio and / or sum and / or difference and / or product between the ratio of and a specific value K can be used as adaptive weight values.

[0899] For example, N s and N a The ratio and / or sum and / or difference and / or product between the ratio of and a specific value K can be used as adaptive weight values.

[0900] For example, N p and N s The ratio and / or sum and / or difference and / or product between the ratio of and a specific value K can be used as adaptive weight values.

[0901] As another example, if there are M filtering blocks in a slice where N is greater than or equal to the threshold T, adaptive weight values ​​can be obtained using a specific value K.

[0902] For example, the ratio and / or sum and / or difference and / or product of N and K can be used as adaptive weight values.

[0903] For example, the ratio and / or sum and / or difference and / or product of M and K can be used as adaptive weight values.

[0904] For example, the ratio between N and M and the ratio and / or sum and / or difference and / or product between K can be used as adaptive weight values.

[0905] For example, N a If T or greater, N p and N s The ratio and / or sum and / or difference and / or product between the ratio of and K can be used as adaptive weight values.

[0906] For example, N p and N s The ratio and / or sum and / or difference and / or product between the ratio of and K can be used as adaptive weight values.

[0907] For example, N s and N a The ratio of K and / or the sum and / or difference and / or product of the ratio between K and the ratio of K can be used as adaptive weight values.

[0908] For example, N p and N a The ratio and / or sum and / or difference and / or product between the ratio of and K can be used as adaptive weight values.

[0909] K and / or T may be values ​​stored in the encoder and / or decoder. N, M, and T may be integers, and K may be a real number.

[0910] As an exemplary embodiment, the adaptive weight value may be derived using at least one of the edge size sum and / or edge size average and / or edge directionality parameter.

[0911] The sum of the edge sizes of the filtering block is E s , the average edge size is E e , the edge direction parameter is E x Assume that, and assume that one of the edge size sum, edge size average, or edge directionality parameter of the filtering block is E.

[0912] For example, the ratio and / or sum and / or difference and / or product of E and a specific value K can be used as adaptive weight values.

[0913] For example, E s and E e The ratio and / or sum and / or difference and / or product between the ratio of and a specific value K can be used as adaptive weight values.

[0914] For example, E e and E x The ratio and / or sum and / or difference and / or product between the ratio of and a specific value K can be used as adaptive weight values.

[0915] For example, E s and E x The ratio and / or sum and / or difference and / or product between the ratio of and a specific value K can be used as adaptive weight values.

[0916] As another example, if there are M filtering blocks in a slice where E is greater than or equal to a threshold T, adaptive weight values ​​can be obtained using a specific value K.

[0917] For example, the ratio and / or sum and / or difference and / or product of E and K can be used as adaptive weight values.

[0918] For example, the ratio and / or sum and / or difference and / or product of M and K can be used as adaptive weight values.

[0919] For example, the ratio of E and M and the ratio and / or sum and / or difference and / or product between K can be used as adaptive weight values.

[0920] For example, E x If is greater than or equal to T, E s and E e The ratio and / or sum and / or difference and / or product between the ratio of and K can be used as adaptive weight values.

[0921] For example, E s and E e The ratio and / or sum and / or difference and / or product between the ratio of and K can be used as adaptive weight values.

[0922] For example, E e and E x The ratio of and / or the ratio and / or sum and / or difference and / or product between K can be used as adaptive weight values.

[0923] For example, E s and E x The ratio and / or sum and / or difference and / or product between the ratio of and K can be used as adaptive weight values.

[0924] Here, K and / or T may be values ​​stored in the encoder and / or decoder. E, M, and T may be integers, and K may be a real number.

[0925] Adaptive weight values ​​can be derived based on the feature map inside the neural network-based filter.

[0926] For example, if the total sum of feature map samples is M, adaptive weight values ​​can be derived from the ratio and / or sum and / or difference and / or product of M and a specific value K. Here, M and K can be real numbers. K can be a value stored in the encoder and / or decoder.

[0927] For example, if the number of samples in a feature map that have values ​​between N and M is L, the adaptive weight values ​​can be obtained by the ratio and / or sum and / or difference and / or product of L and a specific value K. Here, L can be an integer, and N, M, and K can be real numbers. N and / or M and / or K can be values ​​stored in the encoder and / or decoder.

[0928] Information regarding whether block classification is performed can be signaled / encoded / decoded at at least one level among CU, PU, ​​TU, CTU, filtering block, slice, frame, and sequence.

[0929] As an example, as shown in FIG. 24, information regarding whether block classification is performed can be encoded and transmitted as a syntax element bs_block_skip_enabled_flag in the slice header.

[0930] Information on whether block classification is performed can be encoded / decoded or inferred based on at least one of the current slice type, QP, slice QP, temporal id, and gradient value of the filtering block, or any information available during the in-loop filtering encoding and / or decoding process, without being limited thereto.

[0931] For example, information on whether to perform block classification can be transmitted by entropy encoding and / or decoding according to the current image slice type.

[0932] For example, block classification execution information may not be transmitted only when the slice type of the current image is an I-slice. In other words, when it is not an I-slice, information indicating block classification execution can be signaled / encoded / decoded.

[0933] For example, block classification information can be transmitted when the slice type of the current image is at least one of I-slice, B-slice, and P-slice.

[0934] Block classification index information may be at least one of the number of classification indices, the classification indexing method used, the classification index threshold, and the number of classification index thresholds.

[0935] Block classification index information can be encoded and / or decoded at at least one level among CU, PU, ​​TU, CTU, filtering block, slice, frame, and sequence.

[0936] For example, the block classification index information above can be transmitted at the CU and slice unit level.

[0937] For example, block classification index information in a slice header within a bitstream can be entropy encoded or decoded depending on whether block classification is performed.

[0938] For example, block classification index information in a picture parameter set within a bitstream can be entropy encoded or decoded depending on whether block classification is performed.

[0939] i block classification index thresholds T1, T2, ..., T i It can be transmitted by entropy encoding and / or decoding.

[0940] The threshold value can be transmitted at at least one level among CU, PU, ​​TU, CTU, filtering block, slice, frame, and sequence.

[0941] The threshold value can be transmitted via fixed-length encoding.

[0942] The thresholds of all classification indices can be entropy encoded and / or decoded.

[0943] The thresholds of the classification index can be entropy encoded and / or decoded as difference values.

[0944] For example, T1-T2, T1-T3, T1-T i The values ​​can be entropy encoded or decoded.

[0945] For example, T1- T2, T2- T3, T i--1 - T i Values ​​can be entropy encoded or decoded.

[0946] For example, T1-C, T2-C, T i - C values ​​can be transmitted by entropy encoding and / or decoding. Here, C can be an integer.

[0947] For example, as illustrated in FIG. 24, the index threshold can be transmitted as the syntax element bs_partition_threshold_minus2 in the slice header.

[0948] i block classification index thresholds T1, T2, ..., T i It may be stored in advance in the encoding device and the decoding device.

[0949] The above i block classification index thresholds may have at least one different unique number, and the unique number may be entropy encoded and / or decoded.

[0950] Block classification index thresholds T1, T2, ... , T iIt can be obtained using a lookup table.

[0951] For example, the block classification index threshold can use a lookup table consisting of M (=1, 2, ...) integer values. Here, the number of lookup tables can be P, and P can be a positive integer.

[0952] A block classification index threshold lookup table can be generated using at least one of the current slice type, QP, slice QP, temporal id, and gradient value of the filtering block, or any information available during the in-loop filtering encoding and / or decoding process, without being limited thereto.

[0953] P block classification index threshold lookup tables can entropy encode and / or decode.

[0954] For example, P lookup tables can be entropy encoded and / or decoded.

[0955] For example, in P lookup tables, I (I ≤ P) lookup tables can be entropy encoded and / or decoded.

[0956] P block classification index threshold lookup tables may be stored in advance in the encoder and / or decoder.

[0957] Here, the above P lookup tables may each have a different unique number.

[0958] The video encoding device can select I (I ≤ P) lookup tables from P lookup tables. And the number of the I selected lookup tables can be entropy encoded and / or decoded.

[0959] The above block classification index thresholds T1, T2, ..., T iIt can be obtained using an encoding or decoding sequence for at least one of CU, PU, ​​TU, CTU, filtering block, slice, tile, and frame.

[0960] For example, if the block classification method is the number of BS partitions, the threshold T j can be the number of BS partitions of the j-th CTU within a predefined region. Here, the predefined region may refer to a region at a higher level than the CTU. For example, the predefined region may be a slice, tile, or frame.

[0961] For example, if the block classification method is the number of BS samples, the threshold T j can be the number of BS samples of the j-th CTU within a predefined area.

[0962] For example, if the block classification method is the sum of the BS values, the threshold T j can be the sum of the BS of the j-th CTU within a predefined area.

[0963] For example, if the block classification method is the number of BS partitions, the threshold T j may be the number of BS partitions of the j-th filtering block within a predefined region. Here, the predefined region may refer to a region at a higher level than the filtering block. For example, the predefined region may be a CTU, slice, tile, or frame.

[0964] For example, if the block classification method is the number of BS samples, the threshold T j can be the number of BS samples of the j-th filtering block within a predefined area.

[0965] For example, if the block classification method is the sum of the BS values, the threshold T j can be the sum of the BS of the j-th filtering block within a predefined area.

[0966] FIG. 24 is an example diagram of a syntax structure for adaptive neural network-based filtering according to one embodiment.

[0967] As illustrated in FIG. 24, syntax elements for adaptive neural network-based filtering can be signaled at the slice level. However, as previously mentioned, the present invention is not limited thereto, and syntax elements may be signaled at other levels, such as filtering blocks, tiles, pictures, or sequences.

[0968] The syntax element slice_nnlf_unified_mode is information for indicating the neural network-based filtering mode applied to the slice among a number of modes. For example, a first value (e.g., 0) indicates that neural network-based filtering is not applied, i.e., is OFF; a second value (e.g., 1) and a third value (e.g., 2) indicate that neural network-based filtering is controlled at the slice level; and a fourth value (e.g., 3) indicates that neural network-based filtering is controlled at the block level. Here, the modes corresponding to the second and third values ​​can be distinguished according to the QP value used as an input variable to the neural network.

[0969] The syntax element bs_block_skip_enabled_flag may be information indicating whether block classification is performed, that is, whether adaptive neural network-based filtering is performed.

[0970] bs_block_skip_enabled_flag can be signaled / encoded / decoded regardless of slice type.

[0971] Alternatively, bs_block_skip_enabled_flag can be signaled depending on the slice type. For example, as shown in FIG. 24, it can be signaled when the slice type is not an I-slice.

[0972] bs_block_skip_enabled_flag may be signaled when neural network-based filtering mode information directs slice level control (e.g., when slice_nnlf_unified_mode is the second or third value).

[0973] The syntax element bs_partition_threshold_minus2 is information about the threshold for block classification. The threshold can be set as bs_partition_threshold_minus2 + 2. bs_partition_threshold_minus2 can be signaled when block classification is performed, that is, when adaptive neural network-based filtering is performed.

[0974] Meanwhile, the signaling of the syntax element bs_block_skip_enabled_flag can be omitted.

[0975] For example, in some embodiments, when neural network-based filtering is applied, adaptive neural network-based filtering is always performed and information about the threshold can be signaled.

[0976] In some other embodiments, adaptive neural network-based filtering is performed without signaling of bs_block_skip_enabled_flag, depending on at least one of slice type or neural network-based filtering mode information, and information about the threshold can be signaled.

[0977] For example, when the slice type is not an I-slice, adaptive neural network-based filtering is performed and information about the threshold can be signaled.

[0978] For example, when neural network-based filtering mode information directs slice level control, adaptive neural network-based filtering is performed and information about the threshold can be signaled.

[0979]

[0980] In the embodiments, methods may be described based on a flowchart comprising a series of steps or units. The method of the embodiments is not limited to the described order of steps, and some steps may be performed in a different order than described and may be performed simultaneously with other steps. Additionally, the steps described by the flowchart, etc. may not be exclusive. Other steps may be included between the steps described by the flowchart, etc. One or more steps described by the flowchart, etc. may be deleted or omitted.

[0981] The embodiments may include examples of various aspects. While not all possible combinations for representing various aspects can be described, those skilled in the art will recognize that combinations other than those explicitly described are possible. Accordingly, the present invention includes all other substitutions, modifications, and changes falling within the scope of the following claims.

[0982] The embodiments can be implemented in the form of program instructions that can be executed through various computer components and recorded on a computer-readable recording medium.

[0983] A computer-readable recording medium may include a non-transitory computer-readable recording medium. A computer-readable recording medium may include magnetic media such as hard disks, floppy disks, and magnetic tapes; optical recording media such as CD-ROMs and DVDs; magneto-optical media such as floptical disks; and hardware devices specifically configured to store and execute program instructions such as ROM, RAM, and flash memory. Hardware devices may be configured to operate as one or more software modules to perform the processes according to the embodiments, and vice versa.

[0984] A computer-readable recording medium may include program instructions, data files, data structures, etc., either alone or in combination. Program instructions recorded on the computer-readable recording medium may be those specifically designed and configured for the embodiments, or they may be those known and available to those skilled in the art of computer software.

[0985] Program instructions may include machine code, such as that generated by a compiler, and may include high-level language code that can be executed by a computer using an interpreter or the like. Program instructions may also be referred to as computer-executable code or a program. In embodiments, program instructions, computer-executable code, and a program may be interchangeable.

[0986] A computer-readable recording medium may contain information used in the embodiments. For example, a computer-readable recording medium may contain a bitstream, and the bitstream may contain information described in the embodiments. The information described in the embodiments may include syntactic elements. Information described in the embodiments, such as syntactic elements, may be understood as computer-executable code in that it drives an encoding device and a decoder to perform specific processing.

[0987] The bitstream may contain computer-executable code. The computer-executable code may contain information described in embodiments, such as syntactic elements. That is to say, the information described in embodiments, such as syntactic elements, may be regarded as computer-executable code within the bitstream or as part of computer-executable code.

[0988] Although the present invention has been described above with specific details such as specific components, limited embodiments, and drawings, this description is provided only to aid in a more comprehensive understanding of the invention and is not limited to the described embodiments, and a person skilled in the art to which the invention pertains can make various modifications and variations from this description.

[0989] Accordingly, the scope of the present invention should not be limited to the described embodiments, and all variations equivalent to or equivalent to the claims set forth below, as well as the claims themselves, shall be considered to fall within the scope of the concept of the present invention.

Claims

1. In an image encoding method using adaptive neural network-based filtering, A step of deriving a boundary strength (BS) for at least one block boundary within a restored image region; A step of determining whether to perform neural network-based filtering using the above BS; and A step of performing neural network-based filtering on the image region based on the above decision. A video encoding method characterized by including 2. In Paragraph 1, The step of determining whether to perform the above neural network-based filtering is, A step of deriving classification parameters based on non-zero BSs; and A step of determining whether to perform neural network-based filtering by comparing the above classification parameters and threshold values. A video encoding method characterized by including 3. In Paragraph 2, An image encoding method characterized in that the above classification parameter is at least one of the number of partitions separated by a boundary line having the above non-zero BSs, the number of samples constituting the above non-zero BSs, or the total sum of the non-zero BSs.

4. In Paragraph 3, An image encoding method characterized in that the neural network-based filtering is performed when the classification parameter is greater than the threshold value.

5. In Paragraph 1, An image encoding method characterized by including the step of encoding information indicating whether the above-mentioned adaptive neural network-based filtering is allowed at a level of a region higher than the image region.

6. In Paragraph 5, A video encoding method characterized in that the level of the upper region is a level of a slice, tile, frame, or sequence.

7. In Paragraph 2, An image encoding method characterized by including a step of encoding information about the above threshold at a level of a region higher than the image region.

8. In an image decoding method using adaptive neural network-based filtering, A step of deriving a boundary strength (BS) for at least one block boundary within a restored image region; A step of determining whether to perform neural network-based filtering using the above BS; and A step of performing neural network-based filtering on the image region based on the above decision. A video decoding method characterized by including 9. A method for transmitting video data to a decoding device, A step of encoding the above video data to generate a bitstream; and The method includes the step of transmitting the bitstream to the decoding device, The step of generating the above bitstream is, A step of deriving a boundary strength (BS) for at least one block boundary within a restored image region; A step of determining whether to perform neural network-based filtering using the above BS; and A step of performing neural network-based filtering on the image region based on the above decision. A method characterized by including

Images