Multiple neural network model for filtering during video coding

Abstract

An example device for filtering decoded video data includes a memory configured to store video data, and one or more processors implemented in circuitry and configured to decode a picture of the video data, code a value of a syntax element representative of a neural network model to be used to filter a portion of the decoded picture, the value representative of an index to a set of predefined neural network models, the index corresponding to a neural network model in the set of predefined neural network models, and filter the portion of the decoded picture using the neural network model corresponding to the index.

Description

[0001] Citation of relevant applications

[0002] This application claims priority to U.S. Application No. 17 / 448,658, filed September 23, 2021, and U.S. Provisional Patent Application No. 63 / 085,092, filed September 29, 2020, the entire contents of which are incorporated herein by reference. U.S. Application No. 17 / 448,658, filed September 23, 2021, claims the benefit of U.S. Provisional Patent Application No. 63 / 085,092, filed September 29, 2020. Technical Field

[0003] This disclosure relates to video coding, including video encoding and video decoding. Background Technology

[0004] Digital video capabilities can be integrated into a wide variety of devices, including digital televisions, digital direct broadcasting systems, wireless broadcasting systems, personal digital assistants (PDAs), laptops or desktop computers, tablets, e-book readers, digital cameras, digital recording devices, digital media players, video game devices, video game consoles, cellular or satellite radio phones, so-called "smartphones," video conferencing equipment, and video streaming devices. Digital video devices implement video codec technologies, such as those described in standards defined by MPEG-2, MPEG-4, ITU-T H.263, ITU-T H.264 / MPEG-4 Part 10, Advanced Video Codec (AVC), ITU-T H.265 / High-Efficiency Video Codec (HEVC), and extensions to such standards. By implementing these video codec technologies, video devices can more efficiently transmit, receive, encode, decode, and / or store digital video information.

[0005] Video coding and decoding techniques include spatial (intra-picture) prediction and / or temporal (inter-picture) prediction to reduce or eliminate inherent redundancy in video sequences. For block-based video coding and decoding, video strips (e.g., video pictures or portions of video pictures) can be segmented into video blocks, which may also be referred to as codec tree units (CTUs), codec units (CUs), and / or codec nodes. For video blocks in an intra-frame coding (I) strip of a picture, encoding can be performed using spatial prediction relative to reference samples in adjacent blocks within the same picture. For video blocks in an inter-frame coding (P or B) strip of a picture, encoding can be performed using spatial prediction relative to reference samples in adjacent blocks within the same picture or temporal prediction relative to reference samples in other reference pictures. A picture can be called a frame, and a reference picture can be called a reference frame. Summary of the Invention

[0006] Generally, this disclosure describes techniques for filtering decoded (e.g., reconstructed) images that may be distorted. The filtering process can be based on neural network techniques. This filtering process can be used in the context of advanced video codecs such as extensions to or successors of ITU-T H.266 / Voice Universal Coding (VVC) and any other video codec. Specifically, the video encoder can select a neural network model for filtering a portion of the decoded picture. For example, the video encoder can perform rate-distortion optimization (RDO) techniques to determine the neural network model. Alternatively, the video encoder can determine the quantization parameter (QP) of that portion of the decoded picture and determine the neural network model to which that QP is mapped. The video encoder can signal the determined neural network model using the QP itself or a separate value of a syntax element representing an index to a set (or subset of the set) of available neural network models.

[0007] In one example, a method for filtering decoded video data includes: decoding images of the video data; encoding / decoding the value of a syntax element representing a neural network model to be used for filtering a portion of the decoded image, the value representing an index to a predefined set of neural network models corresponding to a neural network model in the predefined set of neural network models; and using the neural network model corresponding to the index to filter that portion of the decoded image.

[0008] In another example, an apparatus for filtering decoded video data includes: a memory configured to store the video data; and one or more processors implemented in a circuit and configured to: decode images of the video data; encode and decode values ​​of syntax elements representing neural network models to be used for filtering a portion of the decoded images, the values ​​representing indices to a predefined set of neural network models corresponding to neural network models in the predefined set of neural network models; and use the neural network model corresponding to the index to filter that portion of the decoded image.

[0009] In another example, a computer-readable storage medium stores instructions that, when executed, cause a processor to: decode images of video data; encode and decode values ​​of syntax elements representing neural network models to be used for filtering a portion of the decoded images, the values ​​representing indices to a predefined set of neural network models corresponding to neural network models in the predefined set of neural network models; and use the neural network model corresponding to the index to filter that portion of the decoded image.

[0010] In another example, an apparatus for filtering decoded video data includes: components for decoding images of the video data; components for encoding and decoding values ​​of syntax elements representing neural network models to be used for filtering a portion of the decoded image, the values ​​representing indices to a predefined set of neural network models corresponding to neural network models in the predefined set of neural network models; and components for filtering that portion of the decoded image using the neural network model corresponding to the index.

[0011] Details of one or more examples are set forth in the accompanying drawings and the following description. Other features, objects, and advantages will become apparent from the specification, drawings, and claims. Attached Figure Description

[0012] Figure 1 This is a block diagram illustrating an example video encoding and decoding system that can perform the techniques of this disclosure.

[0013] Figure 2A and Figure 2B This is a conceptual diagram illustrating an example quadtree binary tree (QTBT) structure and its corresponding codec tree unit (CTU).

[0014] Figure 3 This is a conceptual diagram illustrating a hybrid video codec framework.

[0015] Figure 4 This is a conceptual diagram illustrating a hierarchical prediction structure using a group of pictures (GOP) of size 16.

[0016] Figure 5 This is a conceptual diagram illustrating a neural network-based filter with four layers.

[0017] Figure 6 This is a block diagram illustrating an example video encoder that can perform the techniques of this disclosure.

[0018] Figure 7 This is a block diagram illustrating an example video decoder that can perform the techniques disclosed herein.

[0019] Figure 8 This is a flowchart illustrating an example method for encoding the current block according to the techniques disclosed herein.

[0020] Figure 9 This is a flowchart illustrating an example method for decoding the current block according to the techniques of this disclosure. Detailed Implementation

[0021] Video codec standards include ITU-T H.261, ISO / IEC MPEG-1 Visualization, ITU-T H.262 or ISO / IEC MPEG-2 Visualization, ITU-T H.263, ISO / IEC MPEG-4 Visualization and ITU-T H.264 (also known as ISO / IEC MPEG-4 AVC), High Efficiency Video Codec (HEVC) or ITU-T H.265 (including its range extensions), Multi-View Extension (MV-HEVC), and Scalable Extension (SHVC). Another example video codec standard is Universal Video Codec (VVC) or ITU-T H.266, which was developed by the Joint Video Experts Group (JVET) of the ITU-T Video Coding Experts Group (VCEG) and the ISO / IEC Moving Picture Experts Group (MPEG). Version 1 of the VVC specification, hereinafter referred to as "VVC FDIS", is available at http: / / phenix.int-evry.fr / jvet / doc_end_user / documents / 19_Teleconference / WG 11 / jvet-s 2001-v

[0022] Obtained from 17.zip.

[0023] The techniques disclosed herein are generally directed toward filtering techniques using neural network-based filters. Typically, such filters are trained on large datasets that provide good results for general video data, but may not be optimal for a specific sequence of video data. In contrast, this disclosure describes filtering based on multi-model neural networks, which can produce better results (e.g., in terms of bit rate and distortion) for a specific video sequence.

[0024] Figure 1 This is a block diagram illustrating an example video encoding and decoding system 100 that can perform the techniques of this disclosure. The techniques of this disclosure are generally directed to encoding and / or decoding video data. Generally, video data includes any data used for processing video. Thus, video data can include raw, unencoded video, encoded video, decoded (e.g., reconstructed) video, and video metadata (such as signaling notification data).

[0025] like Figure 1As shown, in this example, system 100 includes a source device 102 that provides encoded video data to be decoded and displayed by destination device 116. Specifically, source device 102 provides video data to destination device 116 via computer-readable medium 110. Source device 102 and destination device 116 can include any of a variety of devices, including desktop computers, laptop computers, mobile devices, tablet computers, set-top boxes, mobile phones (such as smartphones), televisions, cameras, display devices, digital media players, video game consoles, video streaming devices, etc. In some cases, source device 102 and destination device 116 may be equipped for wireless communication and may therefore be referred to as wireless communication devices.

[0026] exist Figure 1 In the example, source device 102 includes a video source 104, memory 106, video encoder 200, and output interface 108. Destination device 116 includes an input interface 122, video decoder 300, memory 120, and display device 118. According to this disclosure, the video encoder 200 of source device 102 and the video decoder 300 of destination device 116 can be configured to apply techniques for filtering using multiple neural network models. Thus, source device 102 represents an example of a video encoding device, while destination device 116 represents an example of a video decoding device. In other examples, the source device and destination device may include other components or arrangements. For example, source device 102 may receive video data from an external video source such as an external camera. Similarly, destination device 116 may interface with an external display device, without including an integrated display device.

[0027] like Figure 1 The system 100 shown is merely an example. Generally, any digital video encoding and / or decoding device can perform filtering techniques using multiple neural network models. The source device 102 and destination device 116 are merely examples of such encoding / decoding devices, where the source device 102 generates encoded / decoded video data for transmission to the destination device 116. In this disclosure, an "encoding / decoding" device is referred to as a device that performs data encoding and / or decoding. Thus, video encoder 200 and video decoder 300 represent examples of encoding / decoding devices, specifically a video encoder and a video decoder, respectively. In some examples, the source device 102 and destination device 116 may operate in a substantially symmetrical manner, such that each of the source device 102 and destination device 116 includes video encoding and decoding components. Therefore, system 100 can support one-way or two-way video transmission between video source device 102 and destination device 116, for example, for video streaming, video playback, video broadcasting, or video telephony.

[0028] Generally, video source 104 represents a video data source (i.e., raw, unencoded video data) and provides a continuous sequence of pictures (also called “frames”) of video data to video encoder 200, which encodes the picture data. Video source 104 of source device 102 may include video capture devices such as cameras, video archives containing previously captured raw video, and / or video feed interfaces that receive video from video content providers. As a further alternative, video source 104 may generate computer graphics-based data as source video, or a combination of live video, archived video, and computer-generated video. In each case, video encoder 200 encodes the captured, pre-captured, or computer-generated video data. Video encoder 200 may rearrange the pictures from the receiving order (sometimes referred to as the “display order”) to an encoding / decoding order for encoding and decoding. Video encoder 200 may generate a bitstream including the encoded video data. Source device 102 may then output the encoded video data via output interface 108 to a computer-readable medium 110 for reception and / or retrieval, for example, by input interface 122 of destination device 116.

[0029] The memory 106 of source device 102 and the memory 120 of destination device 116 represent general-purpose memory. In some examples, memories 106 and 120 may store raw video data, such as raw video from video source 104 and raw decoded video data from video decoder 300. Additionally or alternatively, memories 106 and 120 may store software instructions executable by, for example, video encoder 200 and video decoder 300. Although memories 106 and 120 are shown separately from video encoder 200 and video decoder 300 in this example, it should be understood that video encoder 200 and video decoder 300 may also include internal memory for functionally similar or equivalent purposes. Furthermore, memories 106 and 120 may store, for example, encoded video data output from video encoder 200 and input to video decoder 300. In some examples, portions of memories 106 and 120 may be allocated as one or more video buffers, for example, to store raw decoded and / or encoded video data.

[0030] Computer-readable medium 110 can represent any type of medium or device capable of transmitting encoded video data from source device 102 to destination device 116. In one example, computer-readable medium 110 represents a communication medium enabling source device 102 to transmit encoded video data directly to destination device 116 in real time, for example, via a radio frequency network or a computer-based network. According to a communication standard such as a wireless communication protocol, output interface 108 can modulate the transmitted signal including the encoded video data, and input interface 122 can demodulate the received transmitted signal. The communication medium can include any wireless or wired communication medium, such as radio frequency (RF) spectrum or one or more physical transmission lines. The communication medium can form part of a packet-based network, such as a local area network, a wide area network, or a global network such as the Internet. The communication medium can include a router, switch, base station, or any other equipment that facilitates communication from source device 102 to destination device 116.

[0031] In some examples, source device 102 can output encoded data from output interface 108 to storage device 112. Similarly, destination device 116 can access encoded data from storage device 112 via input interface 122. Storage device 112 can include any of a variety of distributed or locally accessed data storage media, such as hard disks, Blu-ray discs, DVDs, CD-ROMs, flash memory, volatile or non-volatile memory, or any other suitable digital storage medium for storing encoded video data.

[0032] In some examples, source device 102 may output encoded video data to file server 114 or another intermediate storage device that may store the encoded video data generated by source device 102. Destination device 116 may access the video data stored from file server 114 via streaming or download.

[0033] File server 114 can be any type of server device capable of storing encoded video data and sending the encoded video data to destination device 116. File server 114 can represent a web server (e.g., for a website), a server configured to provide file transfer protocol services (such as File Transfer Protocol (FTP) or One-Way File Transfer (FLUTE) protocol), a content delivery network (CDN) device, a Hypertext Transfer Protocol (HTTP) server, a Multimedia Broadcast Multicast Service (MBMS) or Enhanced MBMS (eMBMS) server, and / or a Network Attached Storage (NAS) device. Additionally or alternatively, file server 114 can implement one or more HTTP streaming protocols, such as HTTP-based Dynamic Adaptive Streaming (DASH), HTTP Real-Time Streaming (HLS), Real-Time Streaming Protocol (RTSP), HTTP Dynamic Streaming, etc.

[0034] Destination device 116 can access encoded video data from file server 114 via any standard data connection, including an Internet connection. This can include a wireless channel (e.g., a Wi-Fi connection), a wired connection (e.g., a digital subscriber line (DSL), a cable modem, etc.), or a combination of both suitable for accessing encoded video data stored on file server 114. Input interface 122 can be configured to operate according to any one or more of the various protocols discussed above for retrieving or receiving media data from file server 114, or according to other such protocols for retrieving media data.

[0035] Output interface 108 and input interface 122 may represent a wireless transmitter / receiver, a modem, a wired networking component (e.g., an Ethernet card), a wireless communication component operating according to any of the various IEEE 802.11 standards, or other physical components. In examples where output interface 108 and input interface 122 include wireless components, output interface 108 and input interface 122 may be configured to transmit data such as encoded video data according to cellular communication standards such as 4G, 4G-LTE (Long Term Evolution), LTE Advanced, 5G, or similar standards. In some examples where output interface 108 includes a wireless transmitter, output interface 108 and input interface 122 may be configured according to other wireless standards, such as the IEEE 802.11 specification, the IEEE 802.15 specification (e.g., ZigBee). TM ),Bluetooth TMStandards are used to transmit data such as encoded video data. In some examples, source device 102 and / or destination device 116 may include corresponding system-on-chip (SoC) devices. For example, source device 102 may include an SoC device to perform functions belonging to video encoder 200 and / or output interface 108, and destination device 116 may include an SoC device to perform functions belonging to video decoder 300 and / or input interface 122.

[0036] The technology disclosed herein can be applied to video encoding and decoding that supports any of a variety of multimedia applications, such as over-the-air television broadcasting, cable television transmission, satellite television transmission, internet streaming video transmission such as HTTP-based Dynamic Adaptive Streaming (DASH), digital video encoded onto a data storage medium, decoding digital video stored on a data storage medium, or other applications.

[0037] The input interface 122 of the destination device 116 receives an encoded video bitstream from a computer-readable medium 110 (e.g., a communication medium, storage device 112, file server 114, etc.). The encoded video bitstream may include signaling information defined by the video encoder 200 (which is also used by the video decoder 300), such as syntax elements having values ​​describing the characteristics and / or processing of video blocks or other encoding / decoding units (e.g., stripes, pictures, picture groups, sequences, etc.). The display device 118 displays decoded images of the decoded video data to the user. The display device 118 may represent any of a variety of display devices, such as a liquid crystal display (LCD), a plasma display, an organic light-emitting diode (OLED) display, or another type of display device.

[0038] Although not in Figure 1 As shown, but in some examples, the video encoder 200 and video decoder 300 may each be integrated with the audio encoder and / or audio decoder, and may include appropriate MUX-DEMUX units or other hardware and / or software to process multiplexed streams that include both audio and video in a common data stream. Where applicable, the MUX-DEMUX unit may conform to the ITU H.223 multiplexer protocol or other protocols such as User Datagram Protocol (UDP).

[0039] The video encoder 200 and video decoder 300 can each be implemented as any of a variety of suitable encoder and / or decoder circuits, such as one or more microprocessors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), discrete logic, software, hardware, firmware, or any combination thereof. When the technology is implemented in part in software, the device may store instructions for software in a suitable non-transitory computer-readable medium and use one or more processors to execute those instructions in hardware to perform the technology of this disclosure. Each of the video encoder 200 and video decoder 300 may be included in one or more encoders or decoders, either of which may be integrated as part of a combined encoder / decoder (CODEC) in its respective device. Devices including the video encoder 200 and / or video decoder 300 may include integrated circuits, microprocessors, and / or wireless communication devices such as cellular phones.

[0040] The video encoder 200 and video decoder 300 may operate according to video codec standards such as ITU-T H.265, also known as High Efficiency Video Codec (HEVC) or its extensions such as Multi-View and / or Scalable Video Codec Extensions. Alternatively, the video encoder 200 and video decoder 300 may operate according to other proprietary or industry standards such as Multi-Functional Video Codec (VVC). A draft of the VVC standard is described below: “Multi-Functional Video Codec (Draft 9)” by Bross et al., ITU-TSG 16WP3 and the Joint Video Experts Group (JVET) of ISO / IEC JTC 1 / SC 29 / WG 11, 18th meeting: April 15-24, JVET-R2001-v8 (hereinafter referred to as “VVC Draft 9”). However, the techniques disclosed herein are not limited to any particular codec standard.

[0041] Generally, video encoder 200 and video decoder 300 can perform block-based encoding and decoding of images. The term "block" generally refers to a structure that includes data to be processed (e.g., to be encoded, decoded, or otherwise used in the encoding and / or decoding process). For example, a block may include a two-dimensional sample matrix of luminance and / or chrominance data. Generally, video encoder 200 and video decoder 300 can encode and decode video data represented in YUV (e.g., Y, Cb, Cr) format. That is, video encoder 200 and video decoder 300 can encode and decode luminance and chrominance components, where chrominance components may include both red and blue hue components, rather than encoding and decoding red, green, and blue (RGB) data of the image samples. In some examples, video encoder 200 converts received RGB format data to YUV representation before encoding, and video decoder 300 converts YUV representation to RGB format. Alternatively, preprocessing and postprocessing units (not shown) can perform these conversions.

[0042] This disclosure generally relates to the encoding and decoding (e.g., encoding and decoding) of images, including the process of encoding or decoding data of an image. Similarly, this disclosure may relate to the encoding and decoding of blocks of an image, including the process of encoding or decoding data of the blocks, such as prediction and / or residual encoding and decoding. Encoded video bitstreams generally include a series of values ​​representing encoding and decoding decisions (e.g., encoding / decoding modes) and syntax elements that segment images into blocks. Therefore, references to encoding and decoding images or blocks should generally be understood as encoding and decoding the values ​​of the syntax elements that form images or blocks.

[0043] HEVC defines various blocks, including codec units (CUs), prediction units (PUs), and transform units (TUs). According to HEVC, a video codec (such as a video encoder 200) partitions a codec tree unit (CTU) into CUs according to a quadtree structure. That is, the video codec partitions the CTU and CU into four equal, non-overlapping squares, and each node of the quadtree has zero or four child nodes. Nodes without child nodes can be called "leaf nodes," and the CU of such leaf nodes can include one or more PUs and / or one or more TUs. The video codec can further partition PUs and TUs. For example, in HEVC, a residual quadtree (RQT) represents a partition of a TU. In HEVC, a PU represents inter-frame prediction data, while a TU represents residual data. Intra-frame prediction CUs include intra-frame prediction information, such as intra-frame mode indications.

[0044] As another example, video encoder 200 and video decoder 300 can be configured to operate according to VVC. According to VVC, the video codec (such as video encoder 200) segments the image into multiple codec tree units (CTUs). Video encoder 200 can segment the CTUs according to a tree structure such as a quadtree-binary tree (QTBT) structure or a multi-type tree (MTT) structure. The QTBT structure removes the concept of multiple segmentation types, such as the separation between CUs, PUs, and TUs in HEVC. The QTBT structure includes two levels: a first level segmented according to quadtree segmentation, and a second level segmented according to binary tree segmentation. The root node of the QTBT structure corresponds to a CTU. The leaf nodes of the binary tree correspond to codec units (CUs).

[0045] In the MTT partitioning structure, blocks can be partitioned using quadtree (QT) partitioning, binary tree (BT) partitioning, and one or more types of ternary tree (TT) partitioning (also known as triplet tree (TT)) partitioning. A ternary tree or triplet tree partitioning divides a block into three sub-blocks. In some examples, a ternary tree or triplet tree partitioning divides a block into three sub-blocks without partitioning the original block by a center. The partitioning types in MTT (e.g., QT, BT, and TT) can be symmetric or asymmetric.

[0046] In some examples, the video encoder 200 and the video decoder 300 may use a single QTBT or MTT structure to represent each of the luma and chroma components, while in other examples, the video encoder 200 and the video decoder 300 may use two or more QTBT or MTT structures, such as one QTBT / MTT structure for the luma component and another QTBT / MTT structure for the two chroma components (or two QTBT / MTT structures for the respective chroma components).

[0047] The video encoder 200 and video decoder 300 can be configured to use HEVC-compliant quadtree segmentation, QTBT segmentation, MTT segmentation, or other segmentation structures. For illustrative purposes, the description of the techniques disclosed herein is presented with respect to QTBT segmentation. However, it should be understood that the techniques disclosed herein can also be applied to video codecs configured to use quadtree segmentation or other types of segmentation.

[0048] In some examples, a CTU includes a code-decode tree block (CTB) for luma samples, two corresponding CTBs for chroma samples of an image with three sample arrays, or a CTB for samples of a monochrome image or an image encoded using three separate color planes and a syntax structure for encoding and decoding the samples. For a given value of N, a CTB can be an N×N sample block such that dividing the component into a CTB is a partition. The component can be an array or a single sample of one of the three arrays (luma and two chroma) of an image in 4:2:0, 4:2:2, or 4:4:4 color formats, or an array or a single sample of a monochrome image. In some examples, for some values ​​of M and N, the coded block is an M×N sample block such that dividing the CTB into coded blocks is a partition.

[0049] Blocks (e.g., CTUs or CUs) can be grouped in various ways within an image. As an example, a brick can refer to a rectangular area of ​​a row of CTUs within a specific tile in an image. A tile can be a rectangular area of ​​CTUs within a specific tile column and a specific tile row in an image. A tile column refers to a rectangular area of ​​CTUs with a height equal to the image height and a width specified by a syntax element (e.g., as in the image parameter set). A tile row refers to a rectangular area of ​​CTUs with a height specified by a syntax element (e.g., as in the image parameter set) and a width equal to the image width.

[0050] In some examples, a slice can be divided into multiple bricks, each of which may include one or more CTU rows from that slice. A slice that is not divided into multiple bricks can also be called a brick. However, a brick that is a proper subset of a slice cannot be called a slice.

[0051] The bricks in an image can also be arranged in strips. A strip can be an integer number of bricks in the image, which can be exclusively contained in a single Network Abstraction Layer (NAL) unit. In some examples, a strip consists of multiple complete slices or a continuous sequence of complete bricks consisting of only one slice.

[0052] This disclosure uses "N×N" and "N multiplied by N" interchangeably to refer to the sample size of a block (such as a CU or other video block) in the vertical and horizontal dimensions, for example, 16×16 samples or 16 by 16 samples. Generally, a 16×16 CU will have 16 samples in the vertical direction (y = 16) and 16 samples in the horizontal direction (x = 16). Similarly, an N×N CU generally has N samples in the vertical direction and N samples in the horizontal direction, where N represents a non-negative integer value. Samples in a CU can be arranged in rows and columns. Furthermore, a CU does not necessarily have the same number of samples in the horizontal direction as in the vertical direction. For example, a CU can contain N×M samples, where M is not necessarily equal to N.

[0053] The video encoder 200 encodes video data of the CU (Complex Unit), which represents prediction and / or residual information, as well as other information. The prediction information indicates how the CU will be predicted to form a prediction block for the CU. The residual information generally represents the sample-by-sample difference between the samples of the CU before encoding and the prediction block.

[0054] To predict the Cubic Frame (CU), the video encoder 200 generally forms a prediction block of the CU through inter-frame prediction or intra-frame prediction. Inter-frame prediction generally refers to predicting the CU from data of a previously encoded picture, while intra-frame prediction generally refers to predicting the CU from data of a previously encoded picture within the same picture. To perform inter-frame prediction, the video encoder 200 can use one or more motion vectors to generate prediction blocks. The video encoder 200 can typically perform motion search to identify, for example, a reference block that closely matches the CU in aspects of the difference between the CU and a reference block. The video encoder 200 can use sum of absolute differences (SAD), sum of squared differences (SSD), mean absolute difference (MAD), mean squared difference (MSD), or other such difference calculations to compute a difference metric to determine whether the reference block closely matches the current CU. In some examples, the video encoder 200 can use unidirectional or bidirectional prediction to predict the current CU.

[0055] Some examples of VVC also provide an affine motion compensation mode, which can be viewed as an inter-frame prediction mode. In the affine motion compensation mode, the video encoder 200 can determine two or more motion vectors representing non-translational motion, such as zoom in or out, rotation, perspective motion, or other irregular motion types.

[0056] To perform intra-frame prediction, the video encoder 200 can select an intra-frame prediction mode to generate prediction blocks. Some examples of VVC provide sixty-seven intra-frame prediction modes, including various directional modes as well as planar and DC modes. Generally, the video encoder 200 selects an intra-frame prediction mode that describes the neighboring samples of the current block (e.g., a block of a CU) from which the predicted samples of the current block are predicted. Assuming the video encoder 200 encodes and decodes the CTU and CU in raster scan order (from left to right, from top to bottom), such samples can typically be located above, above left, or to the left of the current block in the same picture as the current block.

[0057] The video encoder 200 encodes data representing the prediction mode of the current block. For example, for inter-frame prediction modes, the video encoder 200 can encode data indicating which of the various available inter-frame prediction modes is used and the corresponding motion information. For unidirectional or bidirectional inter-frame prediction, for example, the video encoder 200 can use Advanced Motion Vector Prediction (AMVP) or merge modes to encode motion vectors. The video encoder 200 can use similar modes to encode motion vectors for affine motion compensation modes.

[0058] After prediction (such as intra-frame or inter-frame prediction of a block), the video encoder 200 can compute residual data for the block. Residual data (such as a residual block) represents the sample-by-sample difference between the block and its predicted block, which is formed using the corresponding prediction mode. The video encoder 200 can apply one or more transforms to the residual block to produce transform data in the transform domain rather than the sample domain. For example, the video encoder 200 can apply a Discrete Cosine Transform (DCT), an integer transform, a wavelet transform, or a conceptually similar transform to the residual video data. Additionally, the video encoder 200 can apply a quadratic transform after a first transform, such as a Mode-Dependent Indivisible Quadratic Transform (MDNSST), a Signal-Dependent Transform, a Karhunen-Loeve Transform (KLT), etc. The video encoder 200 produces transform coefficients after applying one or more transforms.

[0059] As described above, after performing any transformation to produce transform coefficients, the video encoder 200 can perform quantization on the transform coefficients. Quantization generally refers to the process of quantizing transform coefficients to reduce the amount of data used to represent them, thereby providing further compression. By performing the quantization process, the video encoder 200 can reduce the bit depth associated with some or all of the transform coefficients. For example, the video encoder 200 can round an n-bit value down to an m-bit value during quantization, where n is greater than m. In some examples, to perform quantization, the video encoder 200 can perform a bit-right shift of the values ​​to be quantized.

[0060] After quantization, the video encoder 200 can scan the transform coefficients to generate a one-dimensional vector from a two-dimensional matrix containing the quantized transform coefficients. The scan can be designed to place higher-energy (and therefore lower-frequency) coefficients before the vector and lower-energy (and therefore higher-frequency) transform coefficients after the vector. In some examples, the video encoder 200 can utilize a predefined scan order to scan the quantized transform coefficients to produce a serialized vector, and then entropy encode the quantized transform coefficients of the vector. In other examples, the video encoder 200 can perform an adaptive scan. After scanning the quantized transform coefficients to form a one-dimensional vector, the video encoder 200 can entropy encode the one-dimensional vector, for example, according to context-adaptive binary arithmetic encoding / decoding (CABAC). The video encoder 200 can also entropy encode the values ​​of syntax elements, which describe metadata associated with the encoded video data used by the video decoder 300 in decoding the video data.

[0061] To perform CABAC, the video encoder 200 can assign context from within a context model to the symbols to be transmitted. For example, the context could relate to whether the neighboring values ​​of a symbol are zero. Probability determination can be based on the context assigned to the symbols.

[0062] The video encoder 200 can further generate syntax data for the video decoder 300, such as block-based syntax data, image-based syntax data, and sequence-based syntax data, for example, in image headers, block headers, and strip headers, or generate other syntax data, such as sequence parameter sets (SPS), image parameter sets (PPS), or video parameter sets (VPS). The video decoder 300 can similarly decode this syntax data to determine how to decode the corresponding video data.

[0063] In this manner, the video encoder 200 can generate a bitstream including encoded video data, such as syntax elements describing the segmentation of the image into blocks (e.g., CUs) and the prediction and / or residual information of the blocks. Finally, the video decoder 300 can receive the bitstream and decode the encoded video data.

[0064] Generally, the video decoder 300 performs the reverse process of the video encoder 200 to decode the encoded video data of the bitstream. For example, the video decoder 300 can use CABAC to decode the values ​​of the syntax elements of the bitstream in a manner substantially similar to (albeit in reverse) the CABAC encoding process of the video encoder 200. The syntax elements can define segmentation information used to segment the image into CTUs and to segment each CTU according to a corresponding segmentation structure such as a QTBT structure to define the CUs of the CTUs. The syntax elements can further define prediction and residual information for blocks (e.g., CUs) of the video data.

[0065] For example, residual information can be represented by quantization transform coefficients. The video decoder 300 can inversely quantize and inversely transform the quantization transform coefficients of a block to reproduce the residual block of that block. The video decoder 300 uses the prediction mode (intra-frame or inter-frame prediction) notified by signaling and associated prediction information (e.g., motion information for inter-frame prediction) to form a prediction block of the block. The video decoder 300 can then combine the prediction block and the residual block (on a sample-by-sample basis) to reproduce the original block. The video decoder 300 can perform additional processing (such as performing a deblocking process) to reduce visual artifacts along block boundaries.

[0066] Generally, this disclosure may relate to "signaling notification" of certain information, such as syntax elements. The term "signaling notification" can generally refer to communication of values ​​for syntax elements and / or other data used to decode encoded video data. That is, video encoder 200 may signal the values ​​of syntax elements in the bitstream. Generally, signaling notification refers to generating values ​​in the bitstream. As described above, source device 102 may transmit the bitstream to destination device 116 substantially in real time (or non-real time, such as when syntax elements are stored in storage device 112 for later retrieval by destination device 116).

[0067] Figure 2A and Figure 2BThis is a conceptual diagram illustrating an example Quadtree Binary Tree (QTBT) structure 130 and its corresponding Code-to-Code-Unit (CTU) 132. Solid lines represent quadtree partitions, and dashed lines indicate binary tree partitions. In each partition (i.e., non-leaf) node of the binary tree, a signaling flag indicates which partition type (i.e., horizontal or vertical) was used, where in this example, 0 indicates a horizontal partition and 1 indicates a vertical partition. For quadtree partitions, since quadtree nodes divide blocks horizontally and vertically into four sub-blocks of equal size, there is no need to indicate the partition type. Accordingly, the video encoder 200 can encode syntax elements (e.g., partition information) at the region tree level (i.e., solid lines) and the prediction tree level (i.e., dashed lines) of the QTBT structure 130, and the video decoder 300 can decode the above. For the CU represented by the terminal leaf node of the QTBT structure 130, the video encoder 200 can encode video data (such as prediction and transform data), and the video decoder 300 can decode the above.

[0068] Generally speaking, Figure 2B The CTU 132 can be associated with parameters that define the size of the block corresponding to the nodes of the first and second level QTBT structure 130. These parameters may include the CTU size (representing the size of the CTU 132 in the sample), the minimum quadtree size (MinQTSize, representing the minimum allowed size of the quadtree leaf nodes), the maximum binary tree size (MaxBTSize, representing the maximum allowed size of the binary tree root node), the maximum binary tree depth (MaxBTDepth, representing the maximum allowed depth of the binary tree), and the minimum binary tree size (MinBTSize, representing the minimum allowed size of the binary tree leaf nodes).

[0069] The root node of the QTBT structure corresponding to CTU can have four child nodes at the first level of the QTBT structure, and each child node can be partitioned according to quadtree partitioning. That is, the nodes at the first level are leaf nodes (without child nodes) or have four child nodes. The example of QTBT structure 130 represents such a node, which includes a child node and a parent node with solid-line branches. If the node at the first level is not larger than the maximum allowed binary tree root node size (MaxBTSize), the node can be further partitioned by the corresponding binary tree. It is possible to iterate the binary tree partitioning of a node until the partitioned node reaches the minimum allowed binary tree leaf node size (MinBTSize) or the maximum allowed binary tree depth (MaxBTDepth). The example of QTBT structure 130 represents such a node as having dashed-line branches. The binary tree leaf nodes are represented as coding units (CUs), which are used for prediction (e.g., intra-picture or inter-picture prediction) and transformation without any further partitioning. As mentioned above, CUs can also be represented as "video blocks" or "blocks".

[0070] In one example of a QTBT segmentation structure, the CTU size is set to 128×128 (luminance samples and two corresponding 64×64 chrominance samples), MinQTSize is set to 16×16, MaxBTSize is set to 64×64, MinBTSize (for width and height) is set to 4, and MaxBTDepth is set to 4. First, quadtree segmentation is applied to the CTU to generate quadtree leaf nodes. Quadtree leaf nodes can have sizes ranging from 16×16 (i.e., MinQTSize) to 128×128 (i.e., the CTU size). If a quadtree leaf node is 128×128, it will not be further segmented into a binary tree because its size exceeds MaxBTSize (64×64 in this example). Otherwise, the quadtree leaf node can be further segmented into a binary tree. Therefore, the quadtree leaf node is also the root node of the binary tree and has a binary tree depth of 0. Further segmentation is not allowed when the binary tree depth reaches MaxBTDepth (4 in this example). A binary tree node with a width equal to MinBTSize (4 in this example) means that the node is not allowed to be further partitioned vertically (i.e., partitioned by width). Similarly, a binary tree node with a height equal to MinBTSize means that the node is not allowed to be further partitioned horizontally (i.e., partitioned by height). As mentioned above, the leaf nodes of the binary tree are called CUs and are further processed according to the prediction and transformation without further partitioning.

[0071] Figure 3This is a conceptual diagram illustrating a hybrid video codec framework 140. Video codec standards since H.261 have been based on the so-called hybrid video codec principle, which... Figure 3 As shown in the diagram. The term "hybrid" refers to the combination of two means used to reduce redundancy in video signals: prediction 142 and transform encoding / decoding using quantization 144 with prediction residuals. Prediction and transform reduce redundancy in video signals by decorrelation, while quantization reduces their precision, ideally by removing only irrelevant details to reduce the amount of data represented by the transform coefficients. This hybrid video encoding / decoding design principle has also been used in two recent standards, ITU-T H.265 / HEVC and ITU-T H.266 / VVC. (Example...) Figure 3 As shown, a modern hybrid video codec includes block segmentation, prediction 142 including motion compensation or inter-picture prediction and intra-picture prediction, transform / quantization 144 including transform and quantization, entropy encoding / decoding 146, and post / in-loop filtering 148.

[0072] Block segmentation is used to divide an image into smaller blocks for operations in the prediction and transformation processes. Early video codec standards used fixed block sizes, typically 16×16 samples. More recent standards (such as HEVC and VVC) employ tree-based segmentation structures to provide flexible segmentation.

[0073] Motion compensation, or inter-frame prediction, leverages the redundancy present between images (and thus "inter-frame") in a video sequence. Predictions are obtained from one or more previously decoded images (i.e., reference images) based on block-based motion compensation used in all modern video codecs. The corresponding region for generating inter-frame predictions is indicated by motion information including motion vectors and reference image indices.

[0074] Figure 4 This is a conceptual diagram illustrating a hierarchical prediction structure 150 using a group of pictures (GOP) size of 16. In recent video codecs, hierarchical prediction structures within GOPs are applied to improve encoding and decoding efficiency.

[0075] Refer again Figure 3 Intra-picture prediction derives predictions for blocks from spatially adjacent (reference) samples that have already been encoded / decoded, utilizing spatial redundancy present within the picture (and therefore "intra-frame"). Angle prediction, DC prediction, and planar or plane prediction are used in recent video codecs, including AVC, HEVC, and VVC.

[0076] Transform: Hybrid video codec standards apply block transforms to prediction residuals (regardless of whether they originate from inter-picture or intra-picture predictions). Early standards (including H.261, H.262, and H.263) employed Discrete Cosine Transform (DCT). In HEVC and VVC, in addition to DCT, more transform kernels are applied to address different statistical problems in specific video signals.

[0077] Quantization aims to reduce the precision of input values ​​or sets of input values ​​to decrease the amount of data required to represent those values. In hybrid video codecs, quantization is typically applied to individual transformed residual samples (i.e., transform coefficients), resulting in integer coefficient levels. In recent video codec standards, the stride is derived from a so-called quantization parameter (QP) that controls fidelity and bit rate. A larger stride reduces the bit rate but also degrades quality, causing video images to exhibit jamming artifacts and blurred details, for example.

[0078] Context Adaptive Binary Arithmetic Codec (CABAC) is a form of entropy coding that has been used in recent video codecs, such as AVC, HEVC, and VVC, due to its high efficiency.

[0079] Post-loop / in-loop filtering is a filtering process (or a combination of such processes) applied to reconstructed images to reduce coded artifacts. The input to the filtering process is typically the reconstructed image, a combination of the reconstructed residual signal (which includes quantization errors) and the prediction. For example... Figure 3 As shown, the reconstructed image after in-loop filtering is stored and used as a reference for inter-image prediction of subsequent images. Encoding and decoding artifacts are primarily determined by the QP (Quality of Precision), therefore QP information is typically used in the design of the filtering process. In HEVC, the loop filter includes deblocking filtering and Sample Adaptive Offset (SAO) filtering. In the VVC standard, an Adaptive Loop Filter (ALF) is introduced as a third filter. The ALF filtering process is shown below:

[0080]

[0081] Where R(i,j) is the sample set before the filtering process, and R′(i,j) is the sample value after the filtering process. f(k,l) represents the filter coefficients, K(x,y) is the clipping function, and c(k,l) represents the clipping parameter. Variables k and l in... and The values ​​vary between L and L, where L represents the filter length. The clipping function K(x,y) = min(y,max(-y,x)), which corresponds to the function Clip3(-y,y,x). The clipping operation introduces nonlinearity to make ALF more efficient by reducing the influence of adjacent sample values ​​that differ too much from the current sample value. In VVC, the filter parameters can be signaled in the bitstream and can be selected from a predefined set of filters. The ALF filtering process can also be summarized by the following equation:

[0082] R'(i,j)=R(i,j)+ALF_residual_ouput(R)

[0083] Figure 5 This is a conceptual diagram illustrating a four-layer neural network-based filter 160. Various studies have shown that embedding neural networks (NNs) in hybrid video encoding / decoding frameworks can improve compression efficiency. To improve prediction efficiency, neural networks have been introduced into intra-frame prediction and inter-frame prediction modules. NN-based loop filtering has also been a prominent research topic in recent years. In some cases, NN-based filtering processes are applied as post-loop filtering. In this case, the filtering process is only applied to the output image, and the unfiltered image is used as a reference image.

[0084] In addition to existing filters, neural network-based filters, such as deblocking filters, sample adaptive offset (SAO), and / or adaptive loop filtering (ALF), can be applied. NN-based filters can also be applied exclusively, where they are designed to replace all existing filters. Additionally or alternatively, NN-based filters can be designed to complement, enhance, or replace any or all other filters.

[0085] like Figure 5 As shown, NN-based filtering can take the reconstructed samples as input, and the intermediate outputs are residual samples, which are added back to the input to refine the input samples. NN filters can use all color components (e.g., Y, U, and V, or Y, Cb, and Cr, i.e., luminance, blue hue chromaticity, and red hue chromaticity) as input to take advantage of cross-component correlations. Different color components can share the same filters (including network structure and model parameters), or each color component can have its own specific filters.

[0086] The filtering process can also be summarized as follows: R′(i,j)=R(i,j)+NN_filter_residual_output(R). The model structure and parameters of (multiple) NN-based filters can be predefined and stored in the encoder and decoder. Filters can also be signaled in the bitstream.

[0087] This disclosure recognizes that in some cases, predefined filters (e.g., NN-based filters or ALF) are trained on large sets of video and image databases. While the filters may often be optimal, they may not be optimal for a particular distorted sequence. This disclosure also recognizes that since a predefined training image / video database may not represent all possible types of video features, applying filters trained on that given database to sequences with different video features may not provide any objective or subjective benefit, but may instead impair objective or subjective quality.

[0088] According to the technology disclosed herein, Figure 1 The video encoder 200 and video decoder 300 can be configured to perform any or all of the following techniques individually or in any combination. Typically, multiple filtering models can be used. For each target region of the input image, the video encoder 200 and / or video decoder 300 can select one or more of the multiple filtering models to perform filtering. The video encoder 200 and video decoder 300 can implicitly derive the selection of the NN model based on information in the bitstream (e.g., based on the quantization parameter (QP)). Alternatively, the video decoder 300 can derive the selection of the NN model using an index explicitly signaled by the video encoder 200 in the bitstream. In some examples, explicit signaling and implicit derivation can be combined.

[0089] In some examples, the video encoder 200 may select one or more models and signal indices in the bitstream for a chosen model. When a model is selected, it is used to filter the corresponding target region of the input image. When multiple models are selected, they are used in combination to filter the target region of the input image. As an example, the video encoder 200 and the video decoder 300 may apply models to the target region separately, and the outputs of the models involved may be combined into a final output of the target region.

[0090] The switching-off filtering process can be used as a candidate. When "off" is selected for the target region, the video encoder 200 and video decoder 300 do not apply any filters, and the output signal is the same as the input signal.

[0091] The granularity of selection and signaling notification (multiple) models can be designed at different levels. Possible levels for the signaling notification filter model index include video parameter set / sequence parameter set / picture parameter set (VPS / SPS / PPS) level, intra-period level, group of pictures (GOP) level, time layer level within a GOP, picture level, strip level, CTU level, or a grid size of N×N specifically designed for filter signaling. The selection of the filter model signaling level can be fixed; the video encoder 200 can signal the selection as a syntax element to the video decoder 300 in the bitstream; or the video decoder 300 can implicitly derive the selection based on information in the bitstream (e.g., picture resolution, QP, etc.).

[0092] In one example implementation, there are N predefined filter models. At the filter model signaling level, the video encoder 200 selects a model and signals the corresponding index to the video decoder 300 to determine which filter model to use. Possible levels for signaling this element include VPS / SPS / PPS level, intra-frame time period level, group of pictures (GOP) level, picture level, strip level, CTU level, or a grid size of N×N specifically designed for filter signaling. The selection of the filter model signaling level can be fixed, signaled as a syntax element in the bitstream, or implicitly derived based on information in the bitstream (e.g., picture resolution, QP, etc.).

[0093] As another example, video encoder 200 can derive a subset of a predefined model and signal that subset to video decoder 300 in the bitstream as a syntax element in the Sequence Parameter Set (SPS), strip header, picture header, adaptive parameter set (APS), or any other high-level syntax element. The size of the subset is referred to herein as M. M can be any value predefined at a lower level (e.g., strip header, picture header, CTU level, grid level, etc.) or signaled in the bitstream. If M > 1, one of the M candidates is selected and signaled as a syntax element in the bitstream.

[0094] As another example, each filter model can be associated with a QP value, and for each image, the video encoder 200 and video decoder 300 can derive a "model selection QP" and select the model with the QP value that is closest to the current image's "model selection QP". In this case, no additional signaling information is required. The information used to derive the "model selection QP" can include: (multiple) current image QPs, (multiple) reference image QPs, (multiple) QPs of images within the same GOP, (multiple) QPs of images within the same intra-frame time period, etc. Alternatively, the QPs of one or more blocks in a strip, or block-level QPs, can be provided to the NN model to filter the current strip or current block.

[0095] As another example, similar to the example above, each filter model can be associated with a QP value, and the video encoder 200 and video decoder 300 can derive a "model selection QP" for each frame. The video encoder 200 and video decoder 300 can derive a subset of all models based on the "model selection QP". The video encoder 200 can determine the model with the model QP closest to the "model selection QP", select one of these models, and signal to the video decoder 300 with an additional index to select a model from the subset derived from the QP.

[0096] As stated above, in any or all of the various examples discussed above, turning off NN-based modeling can be used as a candidate choice. Turning off can be considered an additional rule candidate for the filter model set, and the signaling for the "off" case can be unified with the signaling for other filter models. As another example, "off" can be considered a special candidate, and the signaling is separate from other filter models.

[0097] The video encoder 200 and video decoder 300 can be configured to apply on / off control for multi-model-based filtering as discussed above. That is, the video encoder 200 and video decoder 300 will only filter the residuals and add the result to the input samples if a given control operation (e.g., a flag) asserts that filtering should be applied. This control signal can be expressed as:

[0098] If (ApplyFilter is true) {R′(i,j)=R(i,j)+(filter_residual_output(R)))},

[0099] Otherwise {R′ (i,j) =R(i,j)}

[0100] ApplyFilter is a control operation that can be determined by the video encoder 200, which can signal the decision to the video decoder 300 in the bitstream with data (e.g., flags or other syntax elements).

[0101] As an example, video encoder 200 can calculate the rate-distortion (RD) cost of applying a filter and compare it to the RD cost of not applying a filter. Based on the result of this comparison, video encoder 200 can determine a control action (e.g., setting the value of a given flag to either 0 or 1, where one state indicates that a filter should be applied and the other state indicates that a filter should not be applied). Video encoder 200 can then signal the given flag in the bitstream. Video decoder 300 can parse the given flag and, based on its value, apply or not apply a filter. Since the trained filter may not be optimal for the entire video sequence or for a given frame or region within the sequence, the granularity at which the control action (flag) is signaled becomes important.

[0102] In some examples, the video encoder 200 can derive the values ​​for filter on / off and signal these values ​​in the bitstream as syntax elements in the Sequence Parameter Set (SPS), strip header, picture parameter set (PPS), picture header, adaptive parameter set (APS), or any other high-level syntax element body. The video decoder 300 can use the value of this syntax element to determine whether to use the filter.

[0103] In some examples, video encoder 200 can derive the value of a filter on / off flag and signal it as a block-level (e.g., CTU-level) syntax element in the bitstream. Video decoder 300 can use the value of this syntax element to determine whether to use the filter.

[0104] In some examples, the video encoder 200 can determine the grid size (for signaling) and the value of the filter on / off flag at the corresponding grid, and signal the value of each grid as a corresponding syntax element in the bitstream. The video decoder 300 can use the values ​​of these syntax elements to determine whether to apply a filter to each element of the grid and to the grid itself.

[0105] Instead of using a fixed grid size for signaling (e.g., where the grid size is always fixed at the CTU size), the video encoder 200 can selectively choose different grid sizes for a given strip based on the RD cost. For example, for a given strip, the video encoder 200 can apply different grid sizes (e.g., an M×N size, where M and N can take values ​​in the range [4, 8, 16, ..., frame size]). The video encoder 200 can calculate the corresponding RD cost for filter on and off conditions. Based on the optimal RD cost, the video encoder 200 can select the grid size and the corresponding filter on / off flag and signal the data representing the grid size and filter on / off flag in the bitstream. The grid size can be predefined at the video encoder 200 and the video decoder 300, for example, as shown in Table 1 below:

[0106] Table 1

[0107]

[0108]

[0109] The video encoder 200 can use unary codes, binary codes, truncated binary codes, or variable-length codes to signal the index value of the grid in the bitstream.

[0110] The video decoder 300 can determine the grid size for signaling based on the resolution of the input sequence. For example, a finer-grained grid (e.g., 8×8) is used for lower-resolution sequences, while a coarser-grained grid (e.g., 128×128) is used for higher-resolution sequences.

[0111] In some examples, the video encoder 200 can determine the filter on / off flag separately for each color component. Therefore, luminance (Y), chrominance (Cb), and chromaticity (Cr) have their own separate control flags.

[0112] In some examples, the filter on / off flag is signaled for only a single component. Other components may share the same flag. For example, the video encoder 200 may signal the control flag only for the luma component, and then the Cb and Cr components may share the same flag. In other examples, the video encoder 200 may signal a combined control flag for the luma component and a flag for the chroma component.

[0113] The video encoder 200 and video decoder 300 can apply separate CABAC contexts to signal flags for individual components (e.g., Y, Cb, and Cr can have separate contexts). Furthermore, the video encoder 200 and video decoder 300 can use spatial / temporal adjacent block flag on / off information to determine the context for signaling flags for the current block. Adjacent blocks can be directly above and / or to the left of the current block, or within a number of blocks.

[0114] The distortion metric used to calculate the RD cost for determining the on / off flag can be SAD (sum of absolute differences), SATD (sum of absolute transformation differences), or any other distortion metric.

[0115] Figure 6 This is a block diagram illustrating an example video encoder 200 capable of performing the techniques of this disclosure. Figure 6 This disclosure is for illustrative purposes and should not be construed as a limitation on the techniques extensively exemplified and described herein. For illustrative purposes, this disclosure describes a video encoder 200 in the context of video codec standards such as the developing ITU-T H.265 / HEVC video codec standard and the VVC video codec standard. However, the techniques of this disclosure are not limited to these video codec standards and are generally applicable to other video coding and decoding standards.

[0116] exist Figure 6 In the example, the video encoder 200 includes a video data memory 230, a mode selection unit 202, a residual generation unit 204, a transform processing unit 206, a quantization unit 208, an inverse quantization unit 210, an inverse transform processing unit 212, a reconstruction unit 214, a filter unit 216, a decoded picture buffer (DPB) 218, and an entropy coding unit 220. Any or all of the video data memory 230, mode selection unit 202, residual generation unit 204, transform processing unit 206, quantization unit 208, inverse quantization unit 210, inverse transform processing unit 212, reconstruction unit 214, filter unit 216, DPB 218, and entropy coding unit 220 can be implemented in one or more processors or in processing circuitry. For example, the units of the video encoder 200 can be implemented as one or more circuit or logic elements as part of hardware circuitry, or as part of a processor, ASIC, or FPGA. Furthermore, the video encoder 200 may include additional or alternative processors or processing circuitry to perform these and other functions.

[0117] The video data storage device 230 can store video data that will be encoded by the components of the video encoder 200. The video encoder 200 can obtain data from, for example, a video source 104 (…). Figure 1The video encoder 200 receives video data stored in video data memory 230. DPB 218 can be used as a reference image memory, storing reference video data for use by the video encoder 200 in predicting subsequent video data. Video data memory 230 and DPB 218 can be formed from any of a variety of storage devices, such as dynamic random access memory (DRAM), including synchronous DRAM (SDRAM), magnetoresistive RAM (MRAM), resistive random access RAM (RRAM), or other types of storage devices. Video data memory 230 and DPB 218 can be provided by the same storage device or separate storage devices. In various examples, video data memory 230 can be placed on-chip along with other components of the video encoder 200, as shown, or off-chip relative to those components.

[0118] In this disclosure, references to video data memory 230 should not be construed as being limited to memory internal to video encoder 200 (unless otherwise stated) or memory external to video encoder 200 (unless otherwise stated). Of course, references to video data memory 230 should be understood as reference memory storing video data (e.g., video data of the current block to be encoded) received by video encoder 200 for encoding. Figure 1 The memory 106 can also provide temporary storage for the outputs from the various units of the video encoder 200.

[0119] Shown Figure 6 The various units are used to help understand the operations performed by the video encoder 200. These units can be implemented as fixed-function circuits, programmable circuits, or a combination thereof. Fixed-function circuits are circuits that provide a specific function and are pre-programmed for the operations they can perform. Programmable circuits are circuits that can be programmed to perform various tasks and provide flexible functionality in the operations they can perform. For example, a programmable circuit can execute software or firmware that causes the programmable circuit to operate in a manner defined by the instructions of the software or firmware. Fixed-function circuits can execute software instructions (e.g., to receive or output parameters), but the type of operation performed by a fixed-function circuit is typically immutable. In some examples, one or more units may be different circuit blocks (fixed-function or programmable), and in some examples, one or more units may be integrated circuits.

[0120] The video encoder 200 may include an arithmetic logic unit (ALU), an essential function unit (EFU), digital circuitry, analog circuitry, and / or a programmable core, all formed by programmable circuitry. In an example where the operation of the video encoder 200 is performed using software executed by programmable circuitry, memory 106 ( Figure 1The video encoder 200 may store instructions (e.g., object code) of the software received and executed by the video encoder 200, or another memory (not shown) within the video encoder 200 may store such instructions.

[0121] The video data storage unit 230 is configured to store received video data. The video encoder 200 can retrieve images of the video data from the video data storage unit 230 and provide the video data to the residual generation unit 204 and the mode selection unit 202. The video data in the video data storage unit 230 can be the raw video data to be encoded.

[0122] The mode selection unit 202 includes a motion estimation unit 222, a motion compensation unit 224, and an intra-frame prediction unit 226. The mode selection unit 202 may include additional functional units to perform video prediction based on other prediction modes. As an example, the mode selection unit 202 may include a palette unit, an intra-frame block copying unit (which may be part of the motion estimation unit 222 and / or the motion compensation unit 224), an affine unit, a linear model (LM) unit, etc.

[0123] The mode selection unit 202 typically coordinates multiple coding passes to test combinations of coding parameters and derive the rate-distortion value of such combinations. Coding parameters may include CTU segmentation into CUs, the prediction mode of the CUs, the transformation type of the CU residual data, and the quantization parameters of the CU residual data. The mode selection unit 202 can ultimately select a combination of coding parameters that has a better rate-distortion value than other tested combinations.

[0124] The video encoder 200 can segment images retrieved from the video data storage 230 into a series of CTUs and encapsulate one or more CTUs within a strip. The mode selection unit 202 can segment the CTUs of the image according to a tree structure (such as the QTBT structure or quadtree structure of HEVC described above). As mentioned above, the video encoder 200 can form one or more CUs by segmenting CTUs according to a tree structure. Such CUs can also be referred to as "video blocks" or "blocks".

[0125] Generally, mode selection unit 202 also controls its components (e.g., motion estimation unit 222, motion compensation unit 224, and intra-prediction unit 226) to generate predicted blocks for the current block (e.g., the overlapping portion of PU and TU in the current CU or HEVC). For inter-frame prediction of the current block, motion estimation unit 222 may perform a motion search to identify one or more closely matching reference blocks in one or more reference pictures (e.g., one or more previously encoded / decoded pictures stored in DPB 218). Specifically, motion estimation unit 222 may calculate values ​​representing how similar a potential reference block is to the current block based on, for example, sum of absolute differences (SAD), sum of squared differences (SSD), mean absolute difference (MAD), mean squared error (MSD), etc. Motion estimation unit 222 may typically perform these calculations using sample-by-sample differences between the current block and the reference blocks under consideration. Motion estimation unit 222 may identify reference blocks with the lowest values ​​generated by these calculations, indicating the reference block that most closely matches the current block.

[0126] Motion estimation unit 222 can generate one or more motion vectors (MVs), which define the position of a reference block in a reference image relative to a current block in the current image. Motion estimation unit 222 can then provide the motion vectors to motion compensation unit 224. For example, for unidirectional inter-frame prediction, motion estimation unit 222 can provide a single motion vector, while for bidirectional inter-frame prediction, it can provide two motion vectors. Motion compensation unit 224 can then use the motion vectors to generate prediction blocks. For example, motion compensation unit 224 can use the motion vectors to retrieve data from the reference blocks. As another example, if the motion vectors have fractional sample precision, motion compensation unit 224 can interpolate the prediction blocks according to one or more interpolation filters. Furthermore, for bidirectional inter-frame prediction, motion compensation unit 224 can retrieve data from two reference blocks identified by their respective motion vectors and combine the retrieved data (e.g., by sample-by-sample averaging or weighted averaging).

[0127] As another example, for intra-prediction or intra-prediction codec, intra-prediction unit 226 can generate a prediction block based on samples adjacent to the current block. For example, in directional mode, intra-prediction unit 226 can typically mathematically combine the values ​​of adjacent samples and fill these calculated values ​​into the current block along a defined direction to generate a prediction block. As another example, in DC mode, intra-prediction unit 226 can calculate the average of the adjacent samples of the current block and generate a prediction block to include the obtained average for each sample used in the prediction block.

[0128] Mode selection unit 202 provides the prediction block to residual generation unit 204. Residual generation unit 204 receives the raw, uncoded version of the current block from video data memory 230 and the prediction block from mode selection unit 202. Residual generation unit 204 calculates the sample-by-sample difference between the current block and the prediction block. The resulting sample-by-sample difference defines the residual block of the current block. In some examples, residual generation unit 204 may also use Residual Differential Pulse Codec Modulation (RDPCM) to determine the difference between sample values ​​in the residual block to generate the residual block. In some examples, one or more subtractor circuits performing binary subtraction may be used to form residual generation unit 204.

[0129] In the example where mode selection unit 202 divides a CU into PUs, each PU can be associated with a luma prediction unit and a corresponding chroma prediction unit. Video encoder 200 and video decoder 300 can support PUs of various sizes. As mentioned above, the size of a CU can refer to the size of its luma codec block, and the size of a PU can refer to the size of the luma prediction unit of the PU. Assuming a specific CU has a size of 2N×2N, video encoder 200 can support PU sizes of 2N×2N or N×N for intra-frame prediction, and symmetrical PU sizes of 2N×2N, 2N×N, N×2N, N×N, or similar for inter-frame prediction. Video encoder 200 and video decoder 300 can also support asymmetric segmentation of PU sizes of 2N×nU, 2N×nD, nL×2N, and nR×2N for inter-frame prediction.

[0130] In the example where mode selection unit 202 does not further divide the CU into PUs, each CU can be associated with a luma codec block and a corresponding chroma codec block. As mentioned above, the size of the CU can refer to the size of the luma codec block of the CU. The video encoder 200 and the video decoder 300 can support CU sizes of 2N×2N, 2N×N, or N×2N.

[0131] For other video codec techniques, such as intra-block copy mode codec, affine mode codec, and linear model (LM) mode codec as examples, mode selection unit 202 generates a prediction block for the current block being encoded via the respective unit associated with the codec technique. In some examples, such as palette mode codec, mode selection unit 202 may not generate a prediction block, but instead may generate syntax elements that indicate how the block should be reconstructed based on the selected palette. In such modes, mode selection unit 202 may provide these syntax elements to entropy coding unit 220 for encoding.

[0132] As described above, the residual generation unit 204 receives video data of the current block and the corresponding prediction block. Then, the residual generation unit 204 generates a residual block for the current block. To generate the residual block, the residual generation unit 204 calculates the sample-by-sample difference between the prediction block and the current block.

[0133] Transform processing unit 206 applies one or more transformations to the residual block to generate a block of transform coefficients (referred to herein as a "transform coefficient block"). Transform processing unit 206 may apply various transformations to the residual block to form the transform coefficient block. For example, transform processing unit 206 may apply a Discrete Cosine Transform (DCT), a direction transformation, a Karhunen-Loeve Transform (KLT), or a conceptually similar transformation to the residual block. In some examples, transform processing unit 206 may perform multiple transformations on the residual block, such as an initial transformation and a secondary transformation such as a rotation transformation. In some examples, transform processing unit 206 does not apply any transformations to the residual block.

[0134] Quantization unit 208 can quantize the transform coefficients in the transform coefficient block to produce a quantized transform coefficient block. Quantization unit 208 can quantize the transform coefficients of the transform coefficient block based on the quantization parameter (QP) value associated with the current block. Video encoder 200 (e.g., via mode selection unit 202) can adjust the degree of quantization applied to the transform coefficient block associated with the current block by adjusting the QP value associated with the CU. Quantization may introduce information loss, and therefore, the quantized transform coefficients may have lower precision than the original transform coefficients produced by transform processing unit 206.

[0135] The inverse quantization unit 210 and the inverse transform processing unit 212 can apply inverse quantization and inverse transform to the quantized transform coefficient block, respectively, to reconstruct the residual block from the transform coefficient block. The reconstruction unit 214 can generate a reconstructed block corresponding to the current block based on the reconstructed residual block and the prediction block generated by the mode selection unit 202 (although potentially with some degree of distortion). For example, the reconstruction unit 214 can add samples from the reconstructed residual block to corresponding samples from the prediction block generated by the mode selection unit 202 to generate the reconstructed block.

[0136] Filter unit 216 can perform one or more filtering operations on the reconstructed blocks. For example, filter unit 216 can perform deblocking operations to reduce block artifacts along the edges of the CU. In some examples, the operations of filter unit 216 can be skipped. Filter unit 216 can be configured to perform various techniques of this disclosure, such as determining one or more neural network models (NN models) 232 to be used for filtering the decoded image and / or whether to apply NN model filtering. Mode selection unit 202 can perform RD calculations using both filtered and unfiltered images to determine the RD cost to determine whether to perform NN model filtering, and then provide entropy encoding unit 220 with data indicating, for example, whether to perform NN model filtering, one or more NN models 232 to be used for the current image or a portion thereof.

[0137] Specifically, according to the techniques of this disclosure, filter unit 216 can determine that NN model 232 comprises a set of available NN models that can be applied to decode a portion of an image. In some examples, filter unit 216 can determine that only a subset of NN models 232 is available for decoding that portion of the image (wherein the subset comprises fewer NN models than the complete set of NN models 232). In such examples, filter unit 216 can provide entropy coding unit 220 with data defining the subset of NN models 232 that can be used to decode that portion of the image. Entropy coding unit 220 can signal the data indicating the subset, for example, in SPS, PPS, APS, strip header, image header, or other high-level syntax elements.

[0138] In some examples, filter unit 216 may select one of N models 232 for a portion of the decoded image. To select one of the N models 232, filter unit 216 may determine the quantization parameters of that portion of the decoded image. Alternatively, mode selection unit 202 may perform a rate-distortion optimization (RDO) process to determine which of the N models 232 produces the best RDO performance and select the one among the N models 232 that produces the best RDO performance. Mode selection unit 202 may then provide a value representing an index to the set (or subset) of available N models 232 corresponding to the determined N model that produces the best RDO performance.

[0139] In some examples, the mode selection unit 202 can also determine whether to enable or disable the use of the NN model 232 for, for example, a specific region of a decoded image, the entire decoded image, a sequence of decoded images, etc. The mode selection unit 202 can provide the entropy coding unit 220 with data indicating whether the use of the NN model 232 is enabled or disabled. The entropy coding unit 220 can also encode data indicating whether the use of the NN model 232 is enabled or disabled, for example, in one or more of the following: Video Parameter Set (VPS), Sequence Parameter Set (SPS), Picture Parameter Set (PPS), Picture Header, Strip Header, Adaptive Parameter Set (APS), Intra-Frame Time Segment Level, Picture Group (GOP) Level, Temporal Layer Level within a GOP, Picture Level, Strip Level, Codec Tree Unit (CTU) Level, or Grid Size Level.

[0140] In some examples, portions of the decoded video data can be elements of a grid. That is, the video encoder 200 can divide the decoded image into a grid of multiple elements, such as elements in rows and columns of the decoded image formed by the grid. The entropy coding unit 220 can also encode the data representing the grid, such as the values ​​of syntax elements representing the number of rows and columns and / or the number of grid elements.

[0141] The video encoder 200 stores reconstructed (and in some cases filtered) blocks in the DPB 218. For example, in an example where the operation of the filter unit 216 is not required, the reconstruction unit 214 may store the reconstructed blocks in the DPB 218. In an example where the operation of the filter unit 216 is required, the filter unit 216 may store the filtered reconstructed blocks in the DPB 218. The motion estimation unit 222 and the motion compensation unit 224 may retrieve a reference image from the DPB 218, which is formed by the reconstructed (and potentially filtered) blocks, to perform inter-frame prediction of blocks in subsequently encoded images. Additionally, the intra-frame prediction unit 226 may use the reconstructed blocks in the DPB 218 of the current image to perform intra-frame prediction of other blocks in the current image.

[0142] Generally, entropy coding unit 220 can entropy-encode syntax elements received from other functional components of video encoder 200. For example, entropy coding unit 220 can entropy-encode quantization transform coefficient blocks from quantization unit 208. As another example, entropy coding unit 220 can entropy-encode predictive syntax elements (e.g., motion information for inter-frame prediction or intra-frame mode information for intra-frame prediction) from mode selection unit 202. Entropy coding unit 220 can perform one or more entropy coding operations on syntax elements of another example of video data to generate entropy-encoded data. For example, entropy coding unit 220 can perform context-adaptive variable-length codec (CAVLC), CABAC, variable-to-variable (V2V) length codec, syntax-based context-adaptive binary arithmetic codec (SBAC), probabilistic interval partitioned entropy (PIPE) codec, exponential-Golomb codec, or another type of entropy coding operation on the data. In some examples, entropy coding unit 220 can operate in a bypass mode, in which syntax elements are not entropy-encoded.

[0143] The video encoder 200 can output a bitstream that includes the syntax elements of entropy coding required to reconstruct blocks of strips or images. Specifically, the entropy coding unit 220 can output this bitstream.

[0144] The above operations are described at the block level. This description should be understood as operations applied to luma and / or chroma codec blocks. As mentioned above, in some examples, the luma and chroma codec blocks are the luma and chroma components of the CU. In some examples, the luma and chroma codec blocks are the luma and chroma components of the PU.

[0145] In some examples, it is not necessary to repeat the operations performed for the luma codec block for the chroma codec block. As an example, it is not necessary to repeat the operations used to identify the motion vector (MV) and reference image for the luma codec block to identify the MV and reference image for the chroma block. Instead, the MV of the luma codec block can be scaled to determine the MV of the chroma block, and the reference image can be the same. As another example, the intra-frame prediction process can be the same for both the luma and chroma codec blocks.

[0146] In this way, Figure 6The video encoder 200 represents an example of an example device for filtering decoded video data, the device including: a memory configured to store video data; and one or more processors implemented in circuitry and configured to: decode images of the video data; encode and decode values ​​of syntax elements representing neural network models to be used for filtering a portion of the decoded images, the values ​​representing indices to a predefined set of neural network models corresponding to neural network models in the predefined set of neural network models; and use the neural network model corresponding to the index to filter that portion of the decoded image.

[0147] Figure 7 This is a block diagram illustrating an example video decoder 300 capable of performing the techniques of this disclosure. Provided Figure 7 This disclosure is for illustrative purposes and not for limiting the extensive examples and techniques described herein. For illustrative purposes, this disclosure describes a video decoder 300 based on VCC and HEVC (ITU-T H.265) technologies. However, the techniques of this disclosure can be implemented by video codec devices configured for other video codec standards.

[0148] exist Figure 7 In the example, the video decoder 300 includes a codec picture buffer (CPB) memory 320, an entropy decoding unit 302, a prediction processing unit 304, an inverse quantization unit 306, an inverse transform processing unit 308, a reconstruction unit 310, a filter unit 312, and a decoded picture buffer (DPB) 314. Any or all of the CPB memory 320, entropy decoding unit 302, prediction processing unit 304, inverse quantization unit 306, inverse transform processing unit 308, reconstruction unit 310, filter unit 312, and DPB 314 can be implemented in one or more processors or processing circuits. For example, units of the video decoder 300 can be implemented as one or more circuit or logic elements as part of hardware circuitry, or as part of a processor, ASIC, or FPGA. Furthermore, the video decoder 300 may include additional or alternative processors or processing circuitry to perform these and other functions.

[0149] The prediction processing unit 304 includes a motion compensation unit 316 and an intra-prediction unit 318. The prediction processing unit 304 may include additional units to perform prediction according to other prediction modes. As an example, the prediction processing unit 304 may include a palette unit, an intra-block copying unit (which may form part of the motion compensation unit 316), an affine unit, a linear model (LM) unit, etc. In other examples, the video decoder 300 may include more, fewer, or different functional components.

[0150] CPB memory 320 can store video data, such as encoded video bitstreams, that will be decoded by components of video decoder 300. For example, it can be stored from computer-readable medium 110 ( Figure 1 The video decoder 300 obtains video data stored in the CPB memory 320. The CPB memory 320 may include a CPB that stores encoded video data (e.g., syntax elements) from the encoded video bitstream. Furthermore, the CPB memory 320 may store video data other than syntax elements of the encoded / decoded picture, such as temporary data representing the output from various units of the video decoder 300. The DPB 314 typically stores decoded pictures that the video decoder 300 may output and / or use as reference video data when decoding subsequent data or pictures of the encoded video bitstream. The CPB memory 320 and DPB 314 may be formed of any of a variety of memory devices, such as dynamic random access memory (DRAM), including synchronous DRAM (SDRAM), magnetoresistive RAM (MRAM), resistive random access RAM (RRAM), or other types of memory devices. The CPB memory 320 and DPB 314 may be provided by the same memory device or separate memory devices. In various examples, the CPB memory 320 may be placed on-chip with other components of the video decoder 300, or off-chip relative to those components.

[0151] Additionally or alternatively, in some examples, the video decoder 300 can be drawn from the memory 120 ( Figure 1 The video data to be encoded and decoded is retrieved from the memory. That is, the memory 120 can store data together with the CPB memory 320 as discussed above. Similarly, when some or all of the functions of the video decoder 300 are implemented in software to be executed by the processing circuitry of the video decoder 300, the memory 120 can store instructions to be executed by the video decoder 300.

[0152] Show Figure 7 The various units shown aid in understanding the operations performed by the video decoder 300. These units can be implemented as fixed-function circuits, programmable circuits, or a combination thereof. Similar to... Figure 6Fixed-function circuits are circuits that provide a specific function and are pre-defined in terms of the operations they can perform. Programmable circuits are circuits that can be programmed to perform various tasks and provide flexible functionality in the operations they can perform. For example, a programmable circuit can execute software or firmware that causes the programmable circuit to operate in a manner defined by the instructions in the software or firmware. Fixed-function circuits can execute software instructions (e.g., to receive or output parameters), but the type of operation performed by a fixed-function circuit is usually immutable. In some examples, one or more units may be different circuit blocks (fixed-function or programmable), and in some examples, one or more units may be integrated circuits.

[0153] The video decoder 300 may include an ALU, an EFU, digital circuitry, analog circuitry, and / or a programmable core formed by programmable circuitry. In an example where the operation of the video decoder 300 is performed by software executed on the programmable circuitry, on-chip or off-chip memory may store instructions (e.g., object code) of the software received and executed by the video decoder 300.

[0154] Entropy decoding unit 302 can receive encoded video data from the CPB and perform entropy decoding on the video data to reproduce the syntax elements. Prediction processing unit 304, inverse quantization unit 306, inverse transform processing unit 308, reconstruction unit 310, and filter unit 312 can generate decoded video data based on the syntax elements extracted from the bitstream.

[0155] Generally, the video decoder 300 reconstructs the image on a block-by-block basis. The video decoder 300 can perform the reconstruction operation on each block individually (where the block currently being reconstructed (i.e., decoded) can be referred to as the "current block").

[0156] Entropy decoding unit 302 can entropy decode the syntax elements of the quantized transform coefficients that define the quantized transform coefficient block, as well as transform information such as quantization parameters (QP) and / or (multiple) transform mode indications. Inverse quantization unit 306 can determine the quantization level using the QP associated with the quantized transform coefficient block, and similarly, determine the inverse quantization level for application by inverse quantization unit 306. Inverse quantization unit 306 can (e.g., by performing a bit-left shift operation) inverse quantize the quantized transform coefficients. Inverse quantization unit 306 can thus form a transform coefficient block including the transform coefficients.

[0157] After the inverse quantization unit 306 forms the transform coefficient block, the inverse transform processing unit 308 can apply one or more inverse transforms to the transform coefficient block to generate a residual block associated with the current block. For example, the inverse transform processing unit 308 can apply the inverse DCT, inverse integer transform, inverse Karhunen-Loeve transform (KLT), inverse rotation transform, inverse direction transform, or another inverse transform to the transform coefficient block.

[0158] Furthermore, the prediction processing unit 304 generates a prediction block based on the prediction information syntax elements entropy decoded by the entropy decoding unit 302. For example, if the prediction information syntax elements indicate that the current block is inter-frame predicted, the motion compensation unit 316 can generate the prediction block. In this case, the prediction information syntax elements can indicate a reference picture in the DPB 314, retrieve a reference block from that reference picture, and indicate a motion vector indicating the position of the reference block in the reference picture relative to the current block in the current picture. The motion compensation unit 316 can typically be configured with the same parameters as the motion compensation unit 224 ( Figure 6 The method described is essentially the same as the method used to perform the inter-frame prediction process.

[0159] As another example, if the prediction information syntax element indicates that the current block is intra-predictable, then intra-prediction unit 318 can generate a prediction block according to the intra-prediction mode indicated by the prediction information syntax element. Again, intra-prediction unit 318 can typically be configured similarly to intra-prediction unit 226. Figure 6 The intra-prediction process is performed in a manner substantially similar to that described above. The intra-prediction unit 318 can retrieve data from neighboring samples of the current block from the DPB 314.

[0160] Reconstruction unit 310 can use prediction blocks and residual blocks to reconstruct the current block. For example, reconstruction unit 310 can add samples from the residual block to corresponding samples from the prediction block to reconstruct the current block.

[0161] Filter unit 312 can perform one or more filtering operations on the reconstructed blocks. For example, filter unit 312 can perform a deblocking operation to reduce block artifacts along the edges of the reconstructed blocks. The operation of filter unit 312 is not necessarily performed in all examples. For example, video decoder 300 can explicitly or implicitly determine whether to use NN models 322 to perform neural network model filtering, for example, using any or all of the various techniques discussed herein. Furthermore, video decoder 300 can explicitly or implicitly determine one or more of the NN models 322 and / or the grid size of the current image to be decoded and filtered. Therefore, when filtering is switched on, filter unit 312 can use one or more of the NN models 322 to filter a portion of the currently decoded image.

[0162] Filter unit 312 can perform one or more filtering operations on the reconstructed blocks. For example, filter unit 312 can perform deblocking operations to reduce block artifacts along the edges of the CU / TU. In some examples, the operations of filter unit 312 can be skipped. Filter unit 312 can be configured to perform various techniques of this disclosure, such as determining one or more neural network models (NN models) 232 to be used for filtering the decoded image and / or whether to apply NN model filtering. Mode selection unit 202 can perform RD calculations using both filtered and unfiltered images to determine the RD cost to determine whether to perform NN model filtering, and then provide entropy encoding unit 220 with data indicating, for example, whether to perform NN model filtering, one or more NN models 322 to be used for the current image or a portion thereof.

[0163] Specifically, according to the techniques of this disclosure, filter unit 312 can determine that NN model 322 includes a set of available NN models that can be applied to decode a portion of an image. In some examples, filter unit 312 can determine that only a subset of NN models 322 is available for decoding that portion of the image (wherein the subset includes fewer NN models than the complete set of NN models 322). In such examples, filter unit 312 can receive decoded data from entropy decoding unit 302 defining the subset of NN models 322 that can be used to decode that portion of the image. Entropy decoding unit 302 can decode data indicating the subset, for example, in SPS, PPS, APS, strip header, image header, or other high-level syntax elements.

[0164] In some examples, filter unit 312 may select one of N models 322 for a portion of the decoded image. To select one of the N models 322, filter unit 312 may determine the quantization parameter (QP) of that portion of the decoded image and determine one of the N models 322 corresponding to the QP. Alternatively, filter unit 312 may receive a value representing an index to a set (or subset) of available N models 322 corresponding to the determined N model to be used.

[0165] In some examples, filter unit 312 can also determine whether to enable or disable the use of NN model 322 for, for example, a specific region of a decoded image, the entire decoded image, a sequence of decoded images, etc. Filter unit 312 can receive data from entropy decoding unit 302 indicating whether the use of NN model 322 is enabled or disabled. Entropy decoding unit 302 can decode data indicating whether the use of NN model 322 is enabled or disabled, for example, in one or more of the following: Video Parameter Set (VPS), Sequence Parameter Set (SPS), Picture Parameter Set (PPS), Picture Header, Strip Header, Adaptive Parameter Set (APS), Intra-Frame Time Segment Level, Picture Group (GOP) Level, Temporal Layer Level within a GOP, Picture Level, Strip Level, Codec Tree Unit (CTU) Level, or Grid Size Level.

[0166] In some examples, the decoded video data may be elements of a grid. That is, the video decoder 300 can divide the decoded image into a grid of multiple elements, such as the elements in the rows and columns of the decoded image formed by the grid. The entropy decoding unit 302 can decode the data representing the grid, such as the values ​​of syntax elements representing the number of rows and columns and / or the number of elements in the grid.

[0167] The video decoder 300 can store reconstructed blocks in the DPB 314. For example, in an example where the filter unit 312 is not operated, the reconstruction unit 310 can store the reconstructed blocks in the DPB 314. In an example where the filter unit 312 is operated, the filter unit 312 can store the filtered reconstructed blocks in the DPB 314. As described above, the DPB 314 can provide reference information to the prediction processing unit 304, such as samples of the current image for intra-frame prediction and previously decoded images for subsequent motion compensation. Furthermore, the video decoder 300 can output the decoded images from the DPB 314 for subsequent rendering in applications such as... Figure 1 The display device 118 is on the display device.

[0168] In this way, Figure 7 The video decoder 300 represents an example of a device for filtering decoded video data, the device including: a memory configured to store video data; and one or more processors implemented in circuitry and configured to: decode images of the video data; encode and decode values ​​of syntax elements representing neural network models to be used for filtering a portion of the decoded images, the values ​​representing indices to a predefined set of neural network models corresponding to neural network models in the predefined set of neural network models; and use the neural network model corresponding to the index to filter a portion of the decoded images.

[0169] Figure 8This is a flowchart illustrating an example method for encoding a current block according to the techniques of this disclosure. The current block may include the current CU. Although this relates to video encoder 200 ( Figure 1 and Figure 3 As described in the description, but it should be understood that other devices can be configured to perform similar actions. Figure 8 The method.

[0170] In this example, the video encoder 200 initially makes a prediction for the current block (350). For example, the video encoder 200 may form a prediction block for the current block. Then, the video encoder 200 may compute a residual block for the current block (352). To compute the residual block, the video encoder 200 may compute the difference between the original, uncoded block and the prediction block for the current block. Then, the video encoder 200 may transform and quantize the coefficients of the residual block (354). Next, the video encoder 200 may scan the quantized transform coefficients of the residual block (356). During or after the scan, the video encoder 200 may entropy encode the coefficients (358). For example, the video encoder 200 may encode the coefficients using CAVLC or CABAC. Then, the video encoder 200 may output the entropy-encoded data of the block (360).

[0171] The video encoder 200 can also decode the current block after encoding it, so that the decoded version of the current block can be used as reference data for subsequently encoded data (e.g., in inter-frame or intra-frame prediction modes). Therefore, the video encoder 200 can perform inverse quantization and inverse transform on the coefficients to reproduce the residual block (362). The video encoder 200 can combine the residual block with the prediction block to form a decoded block (364).

[0172] According to the technology of this disclosure, after encoding and decoding all blocks of an image in the manner described above, the video encoder 200 can determine a neural network (NN) model for a portion of the decoded image including the current block, to be applied to that portion of the decoded image (366). In one example, the video encoder 200 can determine the quantization parameter (QP) of that portion of the decoded image and determine the NN model corresponding to the QP. In another example, the video encoder 200 can perform a rate-distortion optimization (RDO) process to select an NN model, for example, from a set of available NN models. That is, the video encoder 200 can apply various NN models to that portion of the decoded image, then calculate the RDO value of each of the various NN models, and select the NN model that produces the best test RDO value for that portion of the decoded image.

[0173] Then, the video encoder 200 can encode the values ​​representing the NN model (368). When the video encoder 200 selects an NN model based on a QP, the video encoder 200 can simply encode the QP to represent the NN model. That is, for block coding, the QP (e.g., as part of the TU information) can be provided to the NN model to select an appropriate NN-based filter. Alternatively, when the video encoder 200 selects an NN model from a set (or a subset of the set) of available NN models, the video encoder 200 can encode the values ​​of syntax elements representing an index to the set (or subset), where the index corresponds to the position of the NN model in that set or subset. Syntax elements can form a subset of: a video parameter set (VPS), a sequence parameter set (SPS), a picture parameter set (PPS), an adaptive parameter set (APS), an intra-frame time segment level, a picture group (GOP) level, a temporal layer level within a GOP, a picture level, a stripe level, a codec tree unit (CTU) level, or a grid level of a picture grid. Then, the video encoder 200 can apply the determined NN model to that part of the decoded image (369) to filter that part of the decoded image and store the decoded image in the DPB 218.

[0174] In some examples, the video encoder 200 may determine multiple neural network (NN) models for that portion of the decoded image, rather than determining a single NN model. The video encoder 200 may apply each NN model separately to that portion of the decoded image to produce different filtered results. The video encoder 200 may then combine each filtered result to form the final filtered portion of the decoded image.

[0175] This portion of the decoded image can correspond to an element of a grid. That is, the video encoder 200 can divide the decoded image into a grid with multiple elements (e.g., regions formed by a corresponding number of rows and columns). The video encoder 200 can signal data representing the number of rows and columns or the number of elements of the grid in at least one of the following: Video Parameter Set (VPS), Sequence Parameter Set (SPS), Picture Parameter Set (PPS), Picture Header, Strip Header, Adaptive Parameter Set (APS), Intra-Frame Time Segment Level, Picture Group (GOP) Level, Temporal Layer Level within a GOP, Picture Level, Strip Level, Codec Tree Unit (CTU) Level, or Grid Size Level. In some examples, the video encoder 200 can signal values ​​of syntax elements representing indices to possible sets of grid sizes.

[0176] In this way, Figure 8The method represents an example of a method for filtering decoded video data, including: decoding images of the video data; encoding / decoding the value of a syntax element representing a neural network model to be used for filtering a portion of the decoded image, the value representing an index to a predefined set of neural network models, the index corresponding to a neural network model in the predefined set of neural network models; and using the neural network model corresponding to the index to filter that portion of the decoded image.

[0177] Figure 9 This is a flowchart illustrating an example method for decoding the current block according to the techniques of this disclosure. The current block may include the current CU. Although this relates to video decoder 300 ( Figure 1 and Figure 4 This is as described, but it should be understood that other devices can be configured to perform similar actions. Figure 9 The method.

[0178] The video decoder 300 can receive entropy-encoded data of the current block, such as entropy-encoded prediction information and entropy-encoded data of the coefficients of the residual block corresponding to the current block (370). The video decoder 300 can entropy decode the entropy-encoded data to determine the prediction information of the current block, a neural network (NN) model for a portion of the image including the current block, and reproduce the coefficients of the residual block (372). The video decoder 300 can make a prediction for the current block (374), for example, using an intra-frame or inter-frame prediction mode indicated by the prediction information for the current block, to compute a prediction block for the current block. The video decoder 300 can then perform an inverse scan (376) on the reproduced coefficients to create a block of quantized transform coefficients. The video decoder 300 can then perform inverse quantization and inverse transform on the quantized transform coefficients to produce a residual block (378). The video decoder 300 can finally decode the current block by combining the prediction block and the residual block (380).

[0179] The video decoder 300 can also determine from the decoded data (382) the NN model to be applied to a portion of the decoded image including the current block. For example, the video decoder 300 can decode the quantization parameter (QP) for that portion of the decoded image (e.g., for one or more blocks within that portion) and determine the NN model corresponding to that QP. As another example, the video decoder 300 can decode the value of a syntax element representing an index to a set (or subset of a set) of available NN models and determine the NN model corresponding to that index in the set or subset of available NN models. The video decoder 300 can then apply the NN model to that portion of the decoded image (384) to filter that portion of the decoded image. In one example where the filter is an in-loop filter, the video decoder 300 stores the decoded image including the filtered portion in the decoded image buffer 314.

[0180] In this way, Figure 9 The method represents an example of a method for filtering decoded video data, including: decoding images of the video data; encoding / decoding the value of a syntax element representing a neural network model to be used for filtering a portion of the decoded image, the value representing an index to a predefined set of neural network models, the index corresponding to a neural network model in the predefined set of neural network models; and using the neural network model corresponding to the index to filter that portion of the decoded image.

[0181] The following clauses summarize the various techniques of this disclosure:

[0182] Clause 1: A method for filtering decoded video data, the method comprising: determining one or more neural network models to be used for filtering a portion of the decoded image of the video data; and using the one or more neural network models to filter the portion of the decoded image.

[0183] Clause 2: The method according to Clause 1, wherein determining one or more neural network models includes determining no more than a single neural network model.

[0184] Clause 3: The method according to Clause 1, wherein determining one or more neural network models includes determining multiple neural network models.

[0185] Clause 4: According to the method of Clause 3, filtering includes: applying each of the multiple neural network models to the part to form different results; and combining each result to form the final filtered part.

[0186] Clause 5: The method of any one of Clauses 1-4, wherein determining includes using an index of signaling notification in the bitstream including video data.

[0187] Clause 6: The method according to any one of Clauses 1-5, wherein the portion includes elements of a grid.

[0188] Clause 7: The method according to any one of Clauses 1-6 also includes determining the number of elements of the grid.

[0189] Clause 8: According to the method of Clause 7, wherein determining the number of elements of the grid includes decoding at least one of the following syntax elements: Video Parameter Set (VPS), Sequence Parameter Set (SPS), Picture Parameter Set (PPS), Picture Header, Strip Header, Adaptive Parameter Set (APS), Intra-Frame Time Segment Level, Picture Group (GOP) Level, Temporal Layer Level in GOP, Picture Level, Strip Level, Codec Tree Unit (CTU) Level, or Grid Size Level.

[0190] Clause 9: The method according to any one of Clauses 7 and 8, wherein determining the number of elements in the grid includes decoding the value of a syntax element representing an index to a set of possible grid sizes.

[0191] Clause 10: The method according to any one of Clauses 1-9 further includes determining one or more neural network models to be applied before determining one or more neural network models.

[0192] Clause 11: The method according to Clause 10, wherein determining to apply one or more neural network models includes decoding a syntax element having a value indicating that the one or more neural network models are to be applied.

[0193] Clause 12: According to the method of Clause 11, wherein the syntax element is at least one of the following: Video Parameter Set (VPS), Sequence Parameter Set (SPS), Picture Parameter Set (PPS), Picture Header, Strip Header, Adaptive Parameter Set (APS), Intra-Frame Time Segment Level, Picture Group (GOP) Level, Temporal Layer Level in GOP, Picture Level, Strip Level, Codec Tree Unit (CTU) Level, or Grid Size Level.

[0194] Clause 13: The method according to any one of Clauses 1-12, wherein the portion includes decoding a portion of the color components of the image, the color components including one of a luminance component, a blue hue chromaticity component, or a red hue chromaticity component.

[0195] Clause 14: The method according to Clause 13 also includes decoding the syntax elements of the joint representation that filter each of the color components of the decoded image using one or more neural network models.

[0196] Clause 15: The method pursuant to any one of Clauses 1-14 further includes: encoding the current image; and decoding the current image to form the current image.

[0197] Clause 16: The method of Clause 15, wherein determining includes determining based on rate distortion calculations.

[0198] Clause 17: An apparatus for decoding video data, the apparatus comprising one or more components for performing the methods of any one of Clauses 1-16.

[0199] Clause 18: A device pursuant to Clause 17, wherein one or more components include one or more processors implemented in a circuit.

[0200] Clause 19: The device pursuant to Clause 17 also includes a display configured to display decoded video data.

[0201] Clause 20: Devices pursuant to Clause 17, wherein the device includes one or more of a camera, computer, mobile device, broadcast receiver device, or set-top box.

[0202] Clause 21: The device pursuant to Clause 17 also includes a memory configured to store video data.

[0203] Clause 22: A computer-readable storage medium having instructions stored thereon, which, when executed, cause a processor to perform any one of the methods of Clauses 1-16.

[0204] Clause 23: A method for filtering decoded video data, the method comprising: decoding images of the video data; encoding / decoding values ​​of syntax elements representing neural network models to be used for filtering a portion of the decoded images, the values ​​representing indices to a predefined set of neural network models corresponding to neural network models in the predefined set of neural network models; and using the neural network model corresponding to the index to filter that portion of the decoded images.

[0205] Clause 24: The method according to Clause 23 further includes encoding and decoding values ​​representing a subset of a predefined set of neural network models, wherein the neural network model is one of the subsets of the predefined set of neural network models, the subset being smaller than the predefined set of neural network models.

[0206] Clause 25: According to the method of Clause 23, the value of the syntax element representing the neural network model is the value representing multiple neural network models.

[0207] Clause 26: The method according to Clause 25, wherein filtering comprises: applying each of the plurality of neural network models represented by the value to the portion to form different results; and combining each result to form the final filtered portion.

[0208] Clause 27: The method according to Clause 23, wherein encoding or decoding the value of a syntax element includes encoding or decoding the value of the syntax element in one of the following: Video Parameter Set (VPS), Sequence Parameter Set (SPS), Picture Parameter Set (PPS), Adaptive Parameter Set (APS), intra-frame time segment level, Picture Group (GOP) level, temporal layer level in a GOP, picture level, stripe level, codec tree unit (CTU) level, or grid level of a picture grid.

[0209] Clause 28: The method according to Clause 23 also includes segmenting the image according to a grid, wherein the portion includes elements of the grid of the image.

[0210] Clause 29: The method according to Clause 28 also includes determining the number of elements of the grid.

[0211] Clause 30: The method according to Clause 29, wherein determining the number of elements of the grid includes decoding at least one of the following syntax elements: Video Parameter Set (VPS), Sequence Parameter Set (SPS), Picture Parameter Set (PPS), Picture Header, Strip Header, Adaptive Parameter Set (APS), Intra-Frame Time Segment Level, Picture Group (GOP) Level, Temporal Layer Level in GOP, Picture Level, Strip Level, Codec Tree Unit (CTU) Level, or Grid Size Level.

[0212] Clause 31: The method according to Clause 29, wherein determining the number of elements in the grid includes decoding the value of a syntax element representing an index to a set of possible grid sizes.

[0213] Clause 32: The method according to Clause 23 also includes determining the neural network model to be applied before determining the neural network model.

[0214] Clause 33: The method according to Clause 32, wherein determining to apply a neural network model includes decoding the values ​​of the syntax elements indicating to apply the neural network model.

[0215] Clause 34: According to the method of Clause 33, the syntax element is at least one of the following: Video Parameter Set (VPS), Sequence Parameter Set (SPS), Picture Parameter Set (PPS), Picture Header, Strip Header, Adaptive Parameter Set (APS), Intra-Frame Time Segment Level, Picture Group (GOP) Level, Temporal Layer Level in GOP, Picture Level, Strip Level, Codec Tree Unit (CTU) Level, or Grid Size Level.

[0216] Clause 35: The method according to Clause 23, wherein the portion of decoding the image includes a portion of the color components of the image, the color components including one of a luminance component, a blue hue chromaticity component, or a red hue chromaticity component.

[0217] Clause 36: The method according to Clause 35 also includes encoding and decoding the syntax elements of the joint representation that filter each of the color components of the decoded image using a neural network model.

[0218] Clause 37: According to the method of Clause 23, the value of the syntax element includes the quantization parameter (QP) of the portion of the image.

[0219] Clause 38: The method according to Clause 23 also includes encoding the image before decoding the image, wherein encoding / decoding the values ​​of syntax elements includes encoding the values ​​of syntax elements.

[0220] Clause 39: The method pursuant to Clause 38 also includes determining the neural network model based on rate-distortion calculations.

[0221] Clause 40: An apparatus for filtering decoded video data, the apparatus comprising: a memory configured to store video data; and one or more processors implemented in a circuit and configured to: decode images of the video data; encode and decode values ​​of syntax elements representing neural network models to be used for filtering a portion of the decoded images, the values ​​representing indices to a predefined set of neural network models corresponding to neural network models in the predefined set of neural network models; and use the neural network model corresponding to the index to filter the portion of the decoded images.

[0222] Clause 41: A device according to Clause 40, wherein one or more processors are further configured to encode and decode values ​​representing a subset of a predefined set of neural network models, the neural network model being one of a subset of the predefined set of neural network models, the subset being smaller than the predefined set of neural network models.

[0223] Clause 42: A device according to Clause 40, wherein the value of a syntax element representing a neural network model is a value representing a plurality of neural network models, and wherein, in order to filter that portion of a decoded image, one or more processors are configured to: apply each of the plurality of neural network models represented by the value to the portion to form different results; and combine each result to form a final filtered portion.

[0224] Clause 43: A device pursuant to Clause 40, wherein one or more processors are configured to encode or decode the values ​​of syntax elements in one of the following: Video Parameter Set (VPS), Sequence Parameter Set (SPS), Picture Parameter Set (PPS), Adaptive Parameter Set (APS), Intra-Frame Time Segment Level, Picture Group (GOP) Level, Temporal Layer Level within a GOP, Picture Level, Strip Level, Codec Tree Unit (CTU) Level, or Grid Level of a Grid of Pictures.

[0225] Clause 44: A device pursuant to Clause 40, wherein one or more processors are configured to segment an image according to a grid, wherein the segment includes elements of the grid of the image.

[0226] Clause 45: A device pursuant to Clause 44, wherein one or more processors are further configured to decode the value of a syntax element representing the number of elements in a grid, the syntax element comprising at least one of: a video parameter set (VPS), a sequence parameter set (SPS), a picture parameter set (PPS), a picture header, a strip header, an adaptive parameter set (APS), an intra-frame time segment level, a picture group (GOP) level, a time layer level in a GOP, a picture level, a strip level, a codec tree unit (CTU) level, or a grid size level.

[0227] Clause 46: A device according to Clause 40, wherein one or more processors are configured to determine to apply a neural network model before determining a neural network model, wherein, in order to determine to apply a neural network model, one or more processors are configured to decode the values ​​of syntax elements indicating to apply a neural network model.

[0228] Clause 47: A device pursuant to Clause 40, wherein the portion of the decoded image includes a portion of the color components of the decoded image, the color components including one of a luminance component, a blue hue chromaticity component, or a red hue chromaticity component.

[0229] Clause 48: A device pursuant to Clause 40, wherein the value of a syntax element includes the quantization parameter (QP) of a portion of the image.

[0230] Clause 49: A device according to Clause 40, wherein one or more processors are further configured to encode an image prior to decoding the image, and wherein, in order to encode or decode the value of a syntax element, one or more processors are configured to encode the value of a syntax element.

[0231] Clause 50: A device pursuant to Clause 27, wherein one or more processors are further configured to determine a neural network model based on rate distortion calculations.

[0232] Clause 51: The device pursuant to Clause 40 also includes a display configured to display decoded video data.

[0233] Clause 52: Devices pursuant to Clause 40, wherein the device includes one or more of a camera, computer, mobile device, broadcast receiver device, or set-top box.

[0234] Clause 53: A computer-readable storage medium having instructions stored thereon, which, when executed, cause a processor to: decode images of video data; encode and decode values ​​of syntax elements representing neural network models to be used for filtering a portion of the decoded images, the values ​​representing indices to a predefined set of neural network models corresponding to neural network models in the predefined set of neural network models; and use the neural network model corresponding to the index to filter that portion of the decoded images.

[0235] Clause 54: The computer-readable storage medium pursuant to Clause 53 further includes instructions for causing a processor to encode or decode values ​​representing a subset of a predefined set of neural network models, the neural network model being one of a subset of the predefined set of neural network models, the subset being smaller than the predefined set of neural network models.

[0236] Clause 55: A computer-readable storage medium pursuant to Clause 53, wherein the value of a syntax element representing a neural network model is a value representing a plurality of neural network models, and wherein the instructions for causing a processor to filter that portion of a decoded image include instructions for causing the processor to: apply each of the plurality of neural network models represented by the value to the portion to form different results; and combine each result to form a final filtered portion.

[0237] Clause 56: A computer-readable storage medium pursuant to Clause 53, wherein the instructions for causing a processor to encode or decode the values ​​of syntax elements include instructions for causing a processor to encode or decode the values ​​of syntax elements in one of the following: Video Parameter Set (VPS), Sequence Parameter Set (SPS), Picture Parameter Set (PPS), Adaptive Parameter Set (APS), Intra-frame Time Segment Level, Picture Group (GOP) Level, Temporal Layer Level within a GOP, Picture Level, Strip Level, Code-Decoder Tree Unit (CTU) Level, or Grid Level of a Grid of Pictures.

[0238] Clause 57: The computer-readable storage medium pursuant to Clause 53 also includes instructions for causing a processor to segment an image according to a grid, wherein the portion includes elements of the grid of the image.

[0239] Clause 58: The computer-readable storage medium pursuant to Clause 57 further includes instructions for causing a processor to decode the value of a syntax element representing the number of elements in a grid, the syntax element comprising at least one of: a video parameter set (VPS), a sequence parameter set (SPS), a picture parameter set (PPS), a picture header, a strip header, an adaptive parameter set (APS), an intra-frame time segment level, a picture group (GOP) level, a time layer level in a GOP, a picture level, a strip level, a codec tree unit (CTU) level, or a grid size level.

[0240] Clause 59: The computer-readable storage medium pursuant to Clause 53 further includes instructions for causing the processor to determine, prior to determining, the neural network model to be applied, wherein the instructions for causing the processor to determine the neural network model to be applied include instructions for causing the processor to decode values ​​of syntax elements indicating that the neural network model is to be applied.

[0241] Clause 60: A computer-readable storage medium pursuant to Clause 53, wherein the portion of the decoded picture includes a portion of the color components of the decoded picture, the color components including one of a luminance component, a blue hue chromaticity component, or a red hue chromaticity component.

[0242] Clause 61: A computer-readable storage medium pursuant to Clause 53, wherein the value of a syntax element includes a quantization parameter (QP) of a portion of the image.

[0243] Clause 62: The computer-readable storage medium pursuant to Clause 53 further includes instructions for encoding the image before decoding the image, wherein the instructions for encoding / decoding the values ​​of syntax elements include instructions for encoding the values ​​of syntax elements.

[0244] Clause 63: The computer-readable storage medium pursuant to Clause 62 also includes instructions for causing a processor to determine a neural network model based on rate-distortion calculations.

[0245] Clause 64: An apparatus for filtering decoded video data, the apparatus comprising: means for decoding images of the video data; means for encoding and decoding values ​​of syntax elements representing neural network models to be used for filtering a portion of the decoded images, the values ​​representing an index to a predefined set of neural network models corresponding to a neural network model in the predefined set of neural network models; and means for filtering the portion of the decoded images using the neural network model corresponding to the index.

[0246] Clause 65: A method for filtering decoded video data, the method comprising: decoding images of the video data; encoding / decoding values ​​of syntax elements representing neural network models to be used for filtering a portion of the decoded images, the values ​​representing indices in a predefined set of neural network models corresponding to neural network models in the predefined set of neural network models; and using the neural network model corresponding to the index to filter the portion of the decoded images.

[0247] Clause 66: The method according to Clause 65 also includes encoding and decoding values ​​representing a subset of a predefined set of neural network models, wherein the neural network model is one of the subsets of the predefined set of neural network models, the subset being smaller than the predefined set of neural network models.

[0248] Clause 67: The method according to either Clause 65 or Clause 66, wherein the value of the syntax element representing a neural network model is the value representing multiple neural network models.

[0249] Clause 68: The method according to Clause 67, wherein filtering comprises: applying each of the plurality of neural network models represented by the value to the portion to form different results; and combining each result to form the final filtered portion.

[0250] Clause 69: The method according to any one of Clauses 65-68, wherein encoding or decoding the value of a syntax element includes encoding or decoding the value of the syntax element in one of the following: Video Parameter Set (VPS), Sequence Parameter Set (SPS), Picture Parameter Set (PPS), Adaptive Parameter Set (APS), Intra-frame Time Segment Level, Picture Group (GOP) Level, Temporal Layer Level in a GOP, Picture Level, Strip Level, Codec Tree Unit (CTU) Level, or Grid Level of a Picture Grid.

[0251] Clause 70: The method pursuant to any one of Clauses 65-69 further includes segmenting the image according to a grid, wherein the portion includes elements of the grid of the image.

[0252] Clause 71: The method according to Clause 70 also includes determining the number of elements of the grid.

[0253] Clause 72: The method according to Clause 71, wherein determining the number of elements of the grid includes decoding at least one of the following syntax elements: Video Parameter Set (VPS), Sequence Parameter Set (SPS), Picture Parameter Set (PPS), Picture Header, Strip Header, Adaptive Parameter Set (APS), Intra-Frame Time Segment Level, Picture Group (GOP) Level, Temporal Layer Level in GOP, Picture Level, Strip Level, Codec Tree Unit (CTU) Level, or Grid Size Level.

[0254] Clause 73: The method according to Clause 71, wherein determining the number of elements in the grid includes decoding the value of a syntax element representing an index to a set of possible grid sizes.

[0255] Clause 74: The method pursuant to any one of Clauses 65-73 further includes determining the neural network model to be applied prior to determining the neural network model.

[0256] Clause 75: The method according to Clause 74, wherein determining to apply a neural network model includes decoding the values ​​of the syntax elements indicating to apply the neural network model.

[0257] Clause 76: According to the method of Clause 75, the syntax element is at least one of the following: Video Parameter Set (VPS), Sequence Parameter Set (SPS), Picture Parameter Set (PPS), Picture Header, Strip Header, Adaptive Parameter Set (APS), Intra-Frame Time Segment Level, Picture Group (GOP) Level, Temporal Layer Level in GOP, Picture Level, Strip Level, Codec Tree Unit (CTU) Level, or Grid Size Level.

[0258] Clause 77: The method according to any one of Clauses 65-76, wherein the portion of decoding the image includes a portion of the color components of the image, the color components including one of a luminance component, a blue hue chromaticity component, or a red hue chromaticity component.

[0259] Clause 78: The method according to Clause 77 also includes encoding and decoding syntax elements that use a neural network model to filter each of the color components of the decoded image for the joint representation.

[0260] Clause 79: The method according to any one of Clauses 65-78, wherein the value of the syntax element includes the quantization parameter (QP) of the portion of the image.

[0261] Clause 80: The method according to Clause 65 also includes encoding the image before decoding the image, wherein encoding / decoding the values ​​of syntax elements includes encoding the values ​​of syntax elements.

[0262] Clause 81: The method pursuant to Clause 80 also includes determining the neural network model based on rate-distortion calculations.

[0263] Clause 82: An apparatus for filtering decoded video data, the apparatus comprising: a memory configured to store video data; and one or more processors implemented in a circuit and configured to: decode images of the video data; encode and decode values ​​of syntax elements representing neural network models to be used for filtering a portion of the decoded images, the values ​​representing indices to a predefined set of neural network models corresponding to neural network models in the predefined set of neural network models; and use the neural network model corresponding to the index to filter the portion of the decoded images.

[0264] Clause 83: A device pursuant to Clause 82, wherein one or more processors are further configured to encode and decode values ​​representing a subset of a predefined set of neural network models, the neural network model being one of a subset of the predefined set of neural network models, the subset being smaller than the predefined set of neural network models.

[0265] Clause 84: An apparatus pursuant to any one of Clauses 82 and 83, wherein the value of a syntax element representing a neural network model is a value representing a plurality of neural network models, and wherein, in order to filter that portion of a decoded image, one or more processors are configured to: apply each of the plurality of neural network models represented by the value to the portion to form different results; and combine each result to form a final filtered portion.

[0266] Clause 85: The method according to any one of Clauses 82-84, wherein one or more processors are configured to encode or decode the values ​​of syntax elements in one of the following: Video Parameter Set (VPS), Sequence Parameter Set (SPS), Picture Parameter Set (PPS), Adaptive Parameter Set (APS), Intra-frame Time Segment Level, Picture Group (GOP) Level, Temporal Layer Level in a GOP, Picture Level, Strip Level, Codec Tree Unit (CTU) Level, or Grid Level of a Grid of Pictures.

[0267] Clause 86: A device pursuant to any one of Clauses 82-85, wherein one or more processors are configured to segment an image according to a grid, wherein the segment includes elements of the grid of the image.

[0268] Clause 87: A device pursuant to Clause 86, wherein one or more processors are further configured to decode the value of a syntax element representing the number of elements of a grid, the syntax element comprising at least one of: a video parameter set (VPS), a sequence parameter set (SPS), a picture parameter set (PPS), a picture header, a strip header, an adaptive parameter set (APS), an intra-frame time segment level, a picture group (GOP) level, a time layer level within a GOP, a picture level, a strip level, a codec tree unit (CTU) level, or a grid size level.

[0269] Clause 88: An apparatus pursuant to any one of Clauses 82-87, wherein one or more processors are configured to determine to apply a neural network model prior to determining a neural network model, wherein, in order to determine to apply a neural network model, one or more processors are configured to decode the values ​​of syntax elements indicating to apply a neural network model.

[0270] Clause 89: A device pursuant to any one of Clauses 82-88, wherein the portion of the decoded picture includes a portion of the color components of the decoded picture, the color components including one of a luminance component, a blue hue chromaticity component, or a red hue chromaticity component.

[0271] Clause 90: A device pursuant to any one of Clauses 82-89, wherein the value of a syntax element includes a quantization parameter (QP) of a portion of the image.

[0272] Clause 91: A device pursuant to Clause 82, wherein one or more processors are further configured to encode an image prior to decoding the image, and wherein, in order to encode or decode the value of a syntax element, one or more processors are configured to encode the value of a syntax element.

[0273] Clause 92: A device pursuant to Clause 91, wherein one or more processors are further configured to determine a neural network model based on rate distortion calculations.

[0274] Clause 93: The device pursuant to any one of Clauses 82-92 further includes a display configured to display decoded video data.

[0275] Clause 94: Devices pursuant to any one of Clauses 82-93, wherein the device includes one or more of a camera, computer, mobile device, broadcast receiver device or set-top box.

[0276] Clause 95: A computer-readable storage medium having instructions stored thereon, which, when executed, cause a processor to: decode pictures of video data; encode and decode values ​​of syntax elements representing neural network models to be used for filtering a portion of the decoded pictures, the values ​​representing indices to a predefined set of neural network models corresponding to a neural network model in the predefined set of neural network models; and use the neural network model corresponding to the index to filter that portion of the decoded picture.

[0277] Clause 96: The computer-readable storage medium pursuant to Clause 95 further includes instructions for causing a processor to encode or decode values ​​representing a subset of a predefined set of neural network models, the neural network model being one of a subset of the predefined set of neural network models, the subset being smaller than the predefined set of neural network models.

[0278] Clause 97: A computer-readable storage medium pursuant to any one of Clauses 95 and 96, wherein the value of a syntax element representing a neural network model is a value representing a plurality of neural network models, and wherein the instructions for causing a processor to filter that portion of a decoded image include instructions for causing the processor to: apply each of the plurality of neural network models represented by the value to that portion to form different results; and combine each result to form a final filtered portion.

[0279] Clause 98: A computer-readable storage medium pursuant to any one of Clauses 95-97, wherein the instructions for causing a processor to encode or decode the values ​​of a syntax element include instructions for causing a processor to encode or decode the values ​​of a syntax element in one of the following: a video parameter set (VPS), a sequence parameter set (SPS), a picture parameter set (PPS), an adaptive parameter set (APS), an intra-frame time segment level, a picture group (GOP) level, a time layer level in a GOP, a picture level, a stripe level, a code-decode tree unit (CTU) level, or a grid level of a grid of pictures.

[0280] Clause 99: A computer-readable storage medium pursuant to any one of Clauses 95-98 further includes instructions for causing a processor to segment a picture according to a grid, wherein the portion includes elements of a grid of pictures.

[0281] Clause 100: The computer-readable storage medium pursuant to Clause 99 further includes instructions for causing a processor to decode the value of a syntax element representing the number of elements of a grid, the syntax element comprising at least one of: a video parameter set (VPS), a sequence parameter set (SPS), a picture parameter set (PPS), a picture header, a strip header, an adaptive parameter set (APS), an intra-frame time segment level, a picture group (GOP) level, a time layer level within a GOP, a picture level, a strip level, a codec tree unit (CTU) level, or a grid size level.

[0282] Clause 101: A computer-readable storage medium pursuant to any one of Clauses 95-100 further includes instructions for causing a processor to determine, prior to determining, a neural network model to be applied, wherein the instructions for causing the processor to determine the neural network model to be applied include instructions for causing the processor to decode values ​​of syntax elements indicating that a neural network model is to be applied.

[0283] Clause 102: A computer-readable storage medium pursuant to any one of Clauses 95-101, wherein the portion of the decoded picture includes a portion of the color components of the decoded picture, the color components including one of a luminance component, a blue hue chromaticity component, or a red hue chromaticity component.

[0284] Clause 103: A computer-readable storage medium pursuant to any one of Clauses 95-102, wherein the value of a syntax element includes a quantization parameter (QP) of a portion of the image.

[0285] Clause 104: A computer-readable storage medium pursuant to any one of Clauses 95-103 further includes instructions for encoding a picture before decoding the picture, wherein the instructions for encoding / decoding the values ​​of syntax elements include instructions for encoding the values ​​of syntax elements.

[0286] Clause 105: The computer-readable storage medium pursuant to Clause 104 also includes instructions for causing the processor to determine a neural network model based on rate-distortion calculations.

[0287] Clause 106: An apparatus for filtering decoded video data, the apparatus comprising: means for decoding images of the video data; means for encoding and decoding values ​​of syntax elements representing neural network models to be used for filtering a portion of the decoded images, the values ​​representing indices to a predefined set of neural network models corresponding to neural network models in the predefined set of neural network models; and means for filtering the portion of the decoded images using the neural network model corresponding to the index.

[0288] It should be recognized that, based on the examples, certain actions or events of any technique described herein may be performed in a different sequence, and may be added together, combined, or omitted (e.g., not all described actions or events are necessary for technical practice). Furthermore, in some examples, actions or events may be performed concurrently rather than sequentially, for example, through multithreading, interrupt handling, or multiple processors.

[0289] In one or more examples, the described functionality may be implemented in hardware, software, firmware, or any combination thereof. If implemented in software, the functionality may be stored or transmitted as one or more instructions or code on or through a computer-readable medium and executed by a hardware-based processing unit. A computer-readable medium may include a computer-readable storage medium, which corresponds to a tangible medium such as a data storage medium, or a communication medium, including, for example, any medium that facilitates the transfer of a computer program from one place to another according to a communication protocol. In this way, a computer-readable medium may generally correspond to (1) a non-transitory tangible computer-readable storage medium, or (2) a communication medium such as a signal or carrier wave. A data storage medium may be any available medium that can be accessed by one or more computers or one or more processors to retrieve instructions, code, and / or data structures to implement the techniques described in this disclosure. A computer program product may include a computer-readable medium.

[0290] By way of example and not limitation, such computer-readable storage media may include RAM, ROM, EEPROM, CD-ROM or other optical disc storage, disk storage or other magnetic storage devices, flash memory, or any other medium that can be used to store required program code in the form of instructions or data structures and that can be accessed by a computer. Moreover, any connection is appropriately referred to as a computer-readable medium. For example, the definition of medium includes coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies (such as infrared, radio, and microwave) used to send instructions from a website, server, or other remote source. However, it should be understood that computer-readable storage media and data storage media do not include connections, carrier waves, signals, or other transient media, but rather refer to non-transient tangible storage media. As used herein, disks and optical discs include compact optical discs (CDs), laser discs, optical discs, digital versatile discs (DVDs), floppy disks, and Blu-ray discs, wherein disks typically reproduce data magnetically, while optical discs reproduce data optically using lasers. Combinations of the above should also be included within the scope of computer-readable media.

[0291] Instructions can be executed by one or more processors, such as one or more digital signal processors (DSPs), general-purpose microprocessors, application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other equivalent integrated or discrete logic circuits. Therefore, the terms "processor" and "processing circuit" as used herein can refer to any of the foregoing structures or any other structure suitable for implementing the techniques described herein. Additionally, in some aspects, the functionality described herein can be provided within dedicated hardware and / or software modules configured for encoding and decoding, or incorporated into a combined codec. Similarly, the technology can be fully implemented in one or more circuit or logic elements.

[0292] The techniques disclosed herein can be implemented in a variety of devices or apparatuses, including wireless mobile phones, integrated circuits (ICs), or IC sets (e.g., chipsets). Various components, modules, or units are described in this disclosure to emphasize functional aspects of a device configured to perform the disclosed techniques, but they do not necessarily need to be implemented by different hardware units. Rather, as described above, various units can be combined in a codec hardware unit or provided by a collection of interoperable hardware units, including one or more processors as described above, combined with suitable software and / or firmware.

[0293] Various examples have been described. These and other examples are within the scope of the appended claims.

Claims

1. A method for filtering decoded video data, the method comprising: Decode images from video data; The values ​​of syntax elements representing multiple neural network models to be used for filtering a portion of the luminance component of the decoded image are decoded, wherein the multiple neural network models are a predefined set of neural network models; as well as Using the plurality of neural network models corresponding to the value to filter the portion of the luminance component of the decoded image includes: Each of the plurality of neural network models represented by the value is individually applied to the same portion of the luminance component of the decoded image to produce different results; as well as Each of the results is combined to form the final filtered portion of the luminance component of the decoded image.

2. The method according to claim 1 further includes decoding the values ​​representing a subset of the predefined neural network model set, wherein the neural network model among the plurality of neural network models is one of the subsets of the predefined neural network model set, and the subset is smaller than the predefined neural network model set.

3. The method according to claim 1, wherein, Decoding the value of the syntax element includes decoding the value of the syntax element in one of the following ways: Video Parameter Set (VPS), Sequence Parameter Set (SPS), Picture Parameter Set (PPS), Adaptive Parameter Set (APS), Intra-frame Time Segment Level, Picture Group (GOP) Level, Time Layer Level within a GOP, Picture Level, Strip Level, Codec Tree Unit (CTU) Level, or Grid Level of the Picture Grid.

4. The method according to claim 1, further comprising segmenting the image according to a grid, wherein, The portion includes the elements of the grid in the image.

5. The method of claim 4, further comprising determining the number of elements of the grid.

6. The method according to claim 5, wherein, The syntax elements include a first syntax element, and wherein determining the number of elements of the grid includes decoding a second syntax element of at least one of the following: Video Parameter Set (VPS), Sequence Parameter Set (SPS), Picture Parameter Set (PPS), Picture Header, Strip Header, Adaptive Parameter Set (APS), Intra-Frame Time Segment Level, Picture Group (GOP) Level, Time Layer Level within a GOP, Picture Level, Strip Level, Codec Tree Unit (CTU) Level, or Grid Size Level.

7. The method according to claim 5, wherein, The syntax element includes a first syntax element, and wherein determining the number of elements in the grid includes decoding the value of a second syntax element that represents an index to a set of possible grid sizes.

8. The method of claim 1, further comprising determining the neural network model to be applied before determining the neural network model among the plurality of neural network models.

9. The method according to claim 8, wherein, The syntax elements include a first syntax element, and wherein determining to apply the neural network model includes decoding the value of a second syntax element that indicates to apply the neural network model.

10. The method according to claim 9, wherein, The second syntax element is at least one of the following: Video Parameter Set (VPS), Sequence Parameter Set (SPS), Picture Parameter Set (PPS), Picture Header, Strip Header, Adaptive Parameter Set (APS), Intra-Frame Time Segment Level, Picture Group (GOP) Level, Time Layer Level within a GOP, Picture Level, Strip Level, Codec Tree Unit (CTU) Level, or Grid Size Level.

11. The method according to claim 1, wherein, The value of the syntax element includes the quantization parameter QP of the portion of the luminance component of the decoded image.

12. The method of claim 1, further comprising determining the neural network model among the plurality of neural network models based on rate-distortion calculation.

13. An apparatus for filtering decoded video data, the apparatus comprising: The memory is configured to store video data; as well as One or more processors are implemented in the circuit and configured to: Decode images from video data; The values ​​of syntax elements representing multiple neural network models to be used for filtering a portion of the luminance component of the decoded image are decoded, wherein the multiple neural network models are a predefined set of neural network models; as well as The plurality of neural network models corresponding to the value are used to filter the portion of the luminance component of the decoded image, wherein, in order to filter the portion of the decoded image, the one or more processors are configured to: Each of the plurality of neural network models represented by the value is individually applied to the same portion of the luminance component of the decoded image to produce different results; as well as Each of the results is combined to form the final filtered portion of the luminance component of the decoded image.

14. The device according to claim 13, wherein, The one or more processors are further configured to decode values ​​representing a subset of the predefined neural network model set, wherein the neural network model in the plurality of neural network models is one of the subsets of the predefined neural network model set, and the subset is smaller than the predefined neural network model set.

15. The device according to claim 13, wherein, The one or more processors are configured to decode the value of the syntax element in one of the following: Video Parameter Set (VPS), Sequence Parameter Set (SPS), Picture Parameter Set (PPS), Adaptive Parameter Set (APS), Intra-frame Time Segment Level, Picture Group (GOP) Level, Time Layer Level within a GOP, Picture Level, Strip Level, Codec Tree Unit (CTU) Level, or Grid Level of the Picture Grid.

16. The device according to claim 13, wherein, The one or more processors are configured to segment the image according to a grid, wherein the segment includes elements of the grid of the image.

17. The device according to claim 16, wherein, The syntax element includes a first syntax element, and wherein the one or more processors are further configured to decode the value of a second syntax element representing the number of elements of the grid, the syntax element including at least one of: Video Parameter Set (VPS), Sequence Parameter Set (SPS), Picture Parameter Set (PPS), Picture Header, Strip Header, Adaptive Parameter Set (APS), Intra-Frame Time Segment Level, Picture Group (GOP) Level, Time Layer Level within a GOP, Picture Level, Strip Level, Codec Tree Unit (CTU) Level, or Grid Size Level.

18. The device according to claim 13, wherein, The syntax element includes a first syntax element, and wherein the one or more processors are configured to determine the neural network model to be applied before determining the neural network model among the plurality of neural network models, wherein, in order to determine the neural network model to be applied, the one or more processors are configured to decode the value of a second syntax element indicating the application of the neural network model.

19. The device according to claim 13, wherein, The value of the syntax element includes the quantization parameter QP of the portion of the luminance component of the decoded image.

20. The device according to claim 13, wherein, The one or more processors are also configured to determine the neural network model among the plurality of neural network models based on rate-distortion calculations.

21. The device of claim 13, further comprising a display configured to display the decoded video data.

22. The device according to claim 13, wherein, The device includes one or more of a camera, a computer, a mobile device, or a broadcast receiver device.

23. A computer-readable storage medium having instructions stored thereon, the instructions causing a processor, when executed, to: Decode images from video data; The values ​​of syntax elements representing multiple neural network models to be used for filtering a portion of the luminance component of the decoded image are decoded, wherein the multiple neural network models are a predefined set of neural network models; as well as The plurality of neural network models corresponding to the value are used to filter the portion of the luminance component of the decoded image, wherein the instructions for the processor to filter the portion of the decoded image include instructions for causing the processor to perform the following operations: Each of the plurality of neural network models represented by the value is individually applied to the same portion of the luminance component of the decoded image to produce different results; as well as Each of the results is combined to form the final filtered portion of the luminance component of the decoded image.

24. The computer-readable storage medium of claim 23, further comprising instructions for causing the processor to decode values ​​representing a subset of the predefined neural network model set, wherein the neural network model among the plurality of neural network models is one of the subset of the predefined neural network model set, the subset being smaller than the predefined neural network model set.

25. The computer-readable storage medium according to claim 23, wherein, The instructions for causing the processor to decode the value of the syntax element include instructions for causing the processor to decode the value of the syntax element in one of the following: Video Parameter Set (VPS), Sequence Parameter Set (SPS), Picture Parameter Set (PPS), Adaptive Parameter Set (APS), Intra-frame Time Segment Level, Picture Group (GOP) Level, Time Layer Level within a GOP, Picture Level, Strip Level, Codec Tree Unit (CTU) Level, or Grid Level of the Picture Grid.

26. The computer-readable storage medium of claim 23, further comprising instructions for causing the processor to segment the image according to a grid, wherein, The portion includes the elements of the grid in the image.

27. The computer-readable storage medium according to claim 26, wherein, The syntax element includes a first syntax element and further includes instructions for causing the processor to decode the value of a second syntax element representing the number of elements of the grid. The syntax element includes at least one of the following: Video Parameter Set (VPS), Sequence Parameter Set (SPS), Picture Parameter Set (PPS), Picture Header, Strip Header, Adaptive Parameter Set (APS), Intra-Frame Time Segment Level, Picture Group (GOP) Level, Time Layer Level within a GOP, Picture Level, Strip Level, Codec Tree Unit (CTU) Level, or Grid Size Level.

28. The computer-readable storage medium according to claim 23, wherein, The syntax element includes a first syntax element and further includes instructions for causing the processor to determine the neural network model to be applied before determining the neural network model among the plurality of neural network models, wherein the instructions for causing the processor to determine the neural network model to be applied include instructions for causing the processor to decode the value of a second syntax element indicating that the neural network model is to be applied.

29. The computer-readable storage medium according to claim 23, wherein, The value of the syntax element includes the quantization parameter QP of the portion of the luminance component of the decoded image.

30. The computer-readable storage medium of claim 23, further comprising instructions for causing the processor to determine a neural network model among the plurality of neural network models based on rate-distortion calculations.

31. An apparatus for filtering decoded video data, the apparatus comprising: A component used to decode images from video data; A component for decoding the values ​​of syntax elements representing multiple neural network models to be used for filtering a portion of the luminance component of a decoded image, wherein the multiple neural network models are a predefined set of neural network models; as well as A component for filtering the portion of the luminance component of the decoded image using the plurality of neural network models corresponding to the value includes: A component for individually applying each of the plurality of neural network models represented by the value to the same portion of the luminance component of the decoded image to form different results; as well as A component used to combine each of the results to form the final filtered portion of the luminance component of the decoded image.

Images