Cross-component prediction of chroma samples

Abstract

During video encoding or decoding, cross-component prediction can be used to predict chroma samples from concatenated reconstructed luma samples. This prediction can be based on the gradient of the concatenated reconstructed luma samples, the downsampling values of the concatenated reconstructed luma samples, or a combination thereof. An exemplary method includes determining a first value associated with a chroma sample by applying a first gradient pattern to the reconstructed values of a first plurality of luma samples, determining a second value associated with the chroma sample by applying a downsampling filter to the reconstructed values of a second plurality of luma samples, and predicting the chroma sample based on the first value and the second value.

Description

Technical Field

[0001] Cross - reference to Related Applications

[0001] This disclosure claims the benefit of priority to U.S. Provisional Patent Application No. 63 / 350,764, filed on June 9, 2022, U.S. Provisional Patent Application No. 63 / 358,172, filed on July 4, 2022, U.S. Provisional Patent Application No. 63 / 402,944, filed on August 31, 2022, and U.S. Patent Application No. 18 / 325,704, filed on May 30, 2023. All of the above applications are hereby incorporated by reference in their entirety.

[0002] Technical Field

[0002] This disclosure generally relates to video processing, and more specifically, to methods and systems for predicting chroma samples based on collocated luma samples.

Background Art

[0003] Background

[0003] Video is a set of static pictures (or "frames") that capture visual information. To reduce memory and transmission bandwidth, video can be compressed before storage or transmission and restored before display. The compression process is usually referred to as encoding, and the restoration process is usually referred to as decoding. Most commonly, there are various video encoding formats that use standardized video encoding techniques based on prediction, transformation, quantization, entropy encoding, and in - loop filtering. Video encoding standards such as the High Efficiency Video Coding (HEVC / H.265) standard, the Versatile Video Coding (VVC / H.266), and the standard AVS standard, which specify a particular video encoding format, have been developed by standardization organizations. As evolving video encoding techniques are successively adopted by video standards, the encoding efficiency of new video encoding standards becomes even higher.

Summary of the Invention

[0004] Summary

[0004] Embodiments of the present disclosure are derived. In some embodiments, the computer-implemented method determines a first value associated with a chroma sample by applying a first gradient pattern to a reconstructed value of a first plurality of luma samples, determines a second value associated with the chroma sample by applying a downsampling filter to a reconstructed value of a second plurality of luma samples, and predicts the chroma sample based on the first value and the second value.

[0005]

[0005] Embodiments of the present disclosure further provide an apparatus for processing video data. The system includes a memory storing a set of instructions and one or more processors configured to execute the set of instructions to cause the apparatus to determine a first value associated with a chroma sample by applying a first gradient pattern to a reconstructed value of a first plurality of luma samples, determine a second value associated with the chroma sample by applying a downsampling filter to a reconstructed value of a second plurality of luma samples, and predict the chroma sample based on the first value and the second value.

[0006]

[0006] Embodiments of the present disclosure further provide a non-transitory computer-readable medium storing a bitstream of video for processing by a method, the method including determining a first value associated with a chroma sample by applying a first gradient pattern to a reconstructed value of a first plurality of luma samples, determining a second value associated with the chroma sample by applying a downsampling filter to a reconstructed value of a second plurality of luma samples, and predicting the chroma sample based on the first value and the second value.

[0007]

[0007] Embodiments of the present disclosure further provide a non-transitory computer-readable storage medium storing a set of instructions executable by one or more processors of a device to cause the device to start a method for processing video data. The method includes determining a first value associated with a chroma sample by applying a first gradient pattern to a reconstructed value of a first plurality of luma samples, determining a second value associated with the chroma sample by applying a downsampling filter to a reconstructed value of a second plurality of luma samples, and predicting the chroma sample based on the first value and the second value.

[0008]

[0008] Embodiments of the present disclosure further provide a method for processing video data. The method includes predicting a chroma sample from concatenated luma samples associated with the chroma sample, the prediction being based on a non-linear model defining a non-linear relationship between a predicted value of the chroma sample and a value associated with the concatenated luma samples.

[0009]

[0009] Embodiments of the present disclosure further provide an apparatus for processing video data. The system includes a memory storing a set of instructions and one or more processors configured to execute the set of instructions to cause the apparatus to predict a chroma sample from concatenated luma samples associated with the chroma sample, the prediction being based on a non-linear model defining a non-linear relationship between a predicted value of the chroma sample and a value associated with the concatenated luma samples.

[0010]

[0010] Embodiments of the present disclosure further provide a non-transitory computer-readable medium storing a video bitstream for processing by a method including predicting a chroma sample from concatenated luma samples associated with the chroma sample, the prediction being based on a non-linear model defining a non-linear relationship between a predicted value of the chroma sample and a value associated with the concatenated luma samples.

[0011]

[0011] Embodiments of the present disclosure further provide a non-transitory computer-readable storage medium storing a set of instructions executable by one or more processors of a device to cause the device to start a method for processing video data. The method includes predicting a chroma sample from a concatenated luma sample associated with the chroma sample, the prediction being based on a non-linear model that defines a non-linear relationship between a predicted value of the chroma sample and a value associated with the concatenated luma sample.

[0012]

[0012] Embodiments of the present disclosure further provide a computer program product including computer program instructions that enable a computer to execute a method including predicting a chroma sample from a concatenated luma sample associated with the chroma sample, the prediction being based on a non-linear model that defines a non-linear relationship between a predicted value of the chroma sample and a value associated with the concatenated luma sample.

[0013]

[0013] Embodiments of the present disclosure further provide a computer program that enables a computer to execute a method including predicting a chroma sample from a concatenated luma sample associated with the chroma sample, the prediction being based on a non-linear model that defines a non-linear relationship between a predicted value of the chroma sample and a value associated with the concatenated luma sample.

[0014] Brief Description of the Drawings

[0014] Embodiments and various aspects of the present disclosure are illustrated in the following detailed description and the accompanying drawings. The various features shown in the figures are not drawn to scale.

Brief Description of the Drawings

[0015]

Figure 1

[0015] It is a schematic diagram showing the structure of an exemplary video sequence according to some embodiments of the present disclosure.

Figure 2A

[0016] Schematic diagram showing an exemplary encoding process of a hybrid video encoding system according to some embodiments of the present disclosure.

Figure 2B

[0017] Schematic diagram showing another exemplary encoding process of a hybrid video encoding system according to some embodiments of the present disclosure.

Figure 3A

[0018] Schematic diagram showing an exemplary decoding process of a hybrid video encoding system according to some embodiments of the present disclosure.

Figure 3B

[0019] Schematic diagram showing another exemplary decoding process of a hybrid video encoding system according to some embodiments of the present disclosure.

Figure 4

[0020] Block diagram of an exemplary apparatus for encoding or decoding video according to some embodiments of the present disclosure.

Figure 5

[0021] Schematic diagram showing an exemplary method of using adjacent samples to derive parameters of an intersection component model according to some embodiments of the present disclosure.

Figure 6

[0022] Schematic diagram showing another exemplary method of using adjacent samples to derive parameters of an intersection component model according to some embodiments of the present disclosure.

Figure 7

[0023] Schematic diagram showing an exemplary method of using samples on adjacent rows to derive parameters of a gradient model according to some embodiments of the present disclosure.

Figure 8

[0024] Schematic diagram showing another exemplary method of using adjacent samples on adjacent rows to derive parameters of a gradient model according to some embodiments of the present disclosure.

Figure 9

[0025] Flowchart of an exemplary method of predicting chroma samples by using an intersection component non - linear model according to some embodiments of the present disclosure.

Figure 10

[0026] A flowchart of an exemplary method for predicting chroma samples by using a gradient model, according to some embodiments of the present disclosure.

DETAILED DESCRIPTION OF THE INVENTION

[0016] Description of Embodiments

[0027] Here, reference is made in detail to exemplary embodiments, which are illustrated in the accompanying drawings. The following description refers to the accompanying drawings, in which the same reference numerals in different drawings represent the same or similar elements unless otherwise indicated. The implementations shown in the following description of the exemplary embodiments do not represent all implementations in accordance with the present invention. Rather, they are merely examples of apparatuses and methods in accordance with aspects related to the present invention as recited in the appended claims. Specific aspects of the present disclosure are described in more detail below. In case of conflict with terms or definitions incorporated by reference, the terms and definitions provided herein shall prevail.

[0017]

[0028] Embodiments provided by the present disclosure are directed to encoding and decoding video information, and more specifically to a method and system for predicting chroma samples based on one or more concatenated luma samples, such a process being referred to as cross-component prediction throughout the present disclosure. As will be described in detail below, cross-component prediction may employ a cross-component linear model (CCLM), a cross-component non-linear model (CCNLM), a gradient model, or a combination thereof.

[0018]

[0029] The disclosed CCLM, CCNLM, and gradient models are used for encoding or decoding video data. Video is a set of static pictures (or "frames") arranged in a time series for storing visual information. A video capture device (e.g., a camera) can be used to capture and store those pictures in a time series, and a video playback device (e.g., a TV, computer, smartphone, tablet computer, video player, or any end-user terminal with a display function) can be used to display such pictures in a time series. In some applications, the video capture device can also transmit the captured video in real time to a video playback device (e.g., a computer with a monitor) for supervision, holding meetings, or live broadcasts, etc.

[0019]

[0030] In order to reduce the memory space and transmission bandwidth required by such applications, videos can be compressed before being stored and transmitted and restored before being displayed. Compression and restoration can be performed by software or special hardware executed by a processor (e.g., the processor of a general-purpose computer). The module for compression is generally referred to as an "encoder", and the module for restoration is generally referred to as a "decoder". The encoder and decoder can be collectively referred to as a "codec". The encoder and decoder can be implemented as any of various suitable hardware, software, or combinations thereof. For example, the hardware implementation of the encoder and decoder may include circuit mechanisms such as one or more microprocessors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), discrete logic, or any combination thereof. The software implementation of the encoder and decoder may include program code, computer-executable instructions, firmware, or any suitable computer-implemented algorithm or process fixed in a computer-readable medium. Video compression and restoration can be performed by various algorithms or standards such as MPEG-1, MPEG-2, MPEG-4, H.26x, the AVS series, or the like. In some applications, the codec can restore the video from the first encoding standard and recompress the restored video using the second encoding standard. In this case, the codec can be referred to as a "transcoder".

[0020]

[0031] The video encoding process can identify and maintain the useful information that can be used to reconstruct the picture and ignore the information that is not important for reconstruction. If the unimportant information that has been ignored cannot be completely reconstructed, such an encoding process can be referred to as "irreversible". Otherwise, it can be referred to as "reversible". Most encoding processes are irreversible, which is a trade-off for reducing the required memory space and transmission bandwidth.

[0021]

[0032] The useful information of the picture being coded (referred to as the "current picture") includes changes with respect to a reference picture (e.g., a previously coded and reconstructed picture). Such changes can include changes in pixel position, brightness, or color, among which the change in position is the most important. The change in the position of a group of pixels representing an object can reflect the movement of the object between the reference picture and the current picture.

[0022]

[0033] A picture coded without referring to another picture (i.e., the picture is its own reference picture) is referred to as an "I picture". When some or all of the blocks within a picture (e.g., blocks generally referring to a part of a video picture) are predicted using intra prediction or inter prediction with one reference picture (e.g., uni - directional prediction), that picture is referred to as a "P picture". When at least one block within itself is predicted using two reference pictures (e.g., bi - directional prediction), that picture is referred to as a "B picture".

[0023]

[0034] Figures 1, 2A, 2B, 3A, 3B, and 4 show the general aspects of the video encoding / decoding apparatus and process used in the disclosed embodiments. In particular, FIG. 1 shows the structure of an exemplary video sequence 100 according to some embodiments of the present disclosure. The video sequence 100 can be a live video, or a captured and archived video. The video 100 can be a real - world video, a computer - generated video (e.g., a computer - game video), or a combination thereof (e.g., a real - world video with an augmented - reality effect). The video sequence 100 can be input from a video capture device (e.g., a camera), a video archive containing previously captured videos (e.g., a video file stored in a storage device), or a video supply interface (e.g., a video broadcast transceiver) for receiving videos from a video content provider.

[0024]

[0035] As shown in FIG. 1, the video sequence 100 may include a series of pictures temporally arranged along a timeline, including pictures 102, 104, 106, and 108. Pictures 102-106 are consecutive, and there are additional pictures between pictures 106 and 108. In FIG. 1, picture 102 is an I picture, and its reference picture is picture 102 itself. Picture 104 is a P picture, and its reference picture is picture 102 as indicated by the arrow. Picture 106 is a B picture, and its reference pictures are pictures 104 and 108 as indicated by the arrows. In some embodiments, the reference picture of a picture (e.g., picture 104) may not be immediately before or after that picture. For example, the reference picture of picture 104 may be a picture before picture 102. The reference pictures of pictures 102-106 are merely examples, and it should be noted that the present disclosure does not limit the embodiments of the reference picture to the examples shown in FIG. 1.

[0025]

[0036] Typically, due to the computational complexity of such tasks, video codecs do not encode or decode an entire picture at once. Instead, they can divide the picture into basic segments and encode or decode the picture segment by segment. Such basic segments are referred to as basic processing units (「BPU」) in the present disclosure. For example, structure 110 in FIG. 1 shows an exemplary structure of a picture (e.g., any of pictures 102-108) of video sequence 100. In structure 110, the picture is divided into 4×4 basic processing units, and their boundaries are shown as dashed lines. In some embodiments, the basic processing unit may be referred to as a 「macroblock」 in some video coding standards (e.g., the MPEG family, H.261, H.263, or H.264 / AVC) or an 「encoding tree unit」 (「CTU」) in some other video coding standards (e.g., H.265 / HEVC, H.266 / VVC, or AVS). The basic processing unit may have a variable size in the picture, such as 128×128, 64×64, 32×32, 16×16, 4×8, 16×32, or any arbitrary shape and size of pixels. The size and shape of the basic processing unit can be selected based on a balance between coding efficiency and the level of detail to be maintained in the basic processing unit for the picture.

[0026]

[0037] The basic processing unit can be a logical unit that can include a group of different types of video data stored in a computer memory (e.g., a video frame buffer). For example, a basic processing unit of a color picture can include a luma component (Y) representing achromatic luminance information, one or more chroma components (e.g., Cb and Cr) representing color information, and related syntax elements, and the luma and chroma components can have the same size of the basic processing unit. The luma and chroma components may be referred to as 「encoding tree blocks」 (「CTB」) in some video coding standards (e.g., H.265 / HEVC, H.266 / VVC, or AVS). Any operation performed on the basic processing unit can be repeatedly performed on each of its luma and chroma components.

[0027]

[0038] Video coding has multiple operation stages, examples of which are shown in FIGS. 2A-2B and FIGS. 3A-3B. At each stage, the size of the basic processing unit can still be too large for processing, so it can be further divided into segments referred to as "basic processing subunits" in the present disclosure. In some embodiments, the basic processing subunit can be referred to as a "block" in some video coding standards (e.g., the MPEG family, H.261, H.263, H.264 / AVC, or AVS) or a "coding unit" ("CU") in some other video coding standards (e.g., H.265 / HEVC, H.266 / VVC, or AVS). The basic processing subunit can have a size that is the same as or smaller than the basic processing unit. Similar to the basic processing unit, the basic processing subunit is also a logical unit that can include groups of different types of video data (e.g., Y, Cb, Cr, and related syntax elements) stored in computer memory (e.g., a video frame buffer). Any operation performed on the basic processing subunit can be repeatedly performed for each of its luma and chroma components. Note that such division can be performed to further levels as required by the processing. Also note that different stages can divide the basic processing unit in different ways.

[0028]

[0039] For example, in the mode decision stage (an example of which is shown in FIG. 2B), the encoder can determine which prediction mode (e.g., intra-picture prediction or inter-picture prediction) to use for the basic processing unit, but the basic processing unit can be too large to make such a decision. The encoder can divide the basic processing unit into multiple basic processing subunits (e.g., CUs as in the case of H.265 / HEVC, H.266 / VVC, or AVS) and determine the type of prediction for each individual basic processing subunit.

[0029]

[0040] As another example, in the prediction stage (an example thereof is shown in FIGS. 2A to 2B), the coder can perform a prediction operation at the level of a basic processing subunit (e.g., a CU). However, in some cases, the basic processing subunit may still be too large to process. The coder can further divide the basic processing subunit into smaller segments (e.g., called "prediction blocks" or "PBs" in H.265 / HEVC, H.266 / VVC, or AVS), and the prediction operation can be performed at that level.

[0030]

[0041] As another example, in the transform stage (an example thereof is shown in FIGS. 2A to 2B), the coder can perform a transform operation for a residual basic processing subunit (e.g., a CU). However, in some cases, the basic processing subunit may still be too large to process. The coder can further divide the basic processing subunit into smaller segments (e.g., called "transform blocks" or "TBs" in H.265 / HEVC, H.266 / VVC, or AVS), and the transform operation can be performed at that level. Note that the division method of the same basic processing subunit may be different in the prediction stage and the transform stage. For example, in H.265 / HEVC, H.266 / VVC, or AVS, the prediction blocks and transform blocks of the same CU may have different sizes and numbers.

[0031]

[0042] In the structure 110 of FIG. 1, the basic processing unit 112 is further divided into 3×3 basic processing subunits, and their boundaries are shown as dotted lines. Different basic processing units of the same picture can be divided into basic processing subunits in different ways.

[0032]

[0043] In some implementations, to provide parallel processing and error resilience capabilities to video encoding and decoding, a picture can be divided into regions for processing, such that the encoding or decoding process need not depend on information from any other region of the picture with respect to one region of the picture. In other words, each region of the picture can be processed independently. By doing so, the codec can process different regions of the picture in parallel, thus increasing the encoding efficiency. When the data of a region is damaged during processing or lost during network transmission, the codec can also correctly encode or decode other regions of the same picture without depending on the damaged or lost data, thus providing an error resilience capability. In some video encoding standards, a picture can be divided into different types of regions. For example, H.265 / HEVC, H.266 / VVC, and AVS provide two types of regions: "slices" and "tiles". It should also be noted that different pictures of video sequence 100 may have different partitioning schemes for dividing the picture into regions.

[0033]

[0044] For example, in FIG. 1, structure 110 is divided into three regions 114, 116, and 118, and their boundaries are shown as solid lines inside structure 110. Region 114 includes four basic processing units. Each of regions 116 and 118 includes six basic processing units. It should be noted that the basic processing units, basic processing subunits, and regions of structure 110 in FIG. 1 are merely examples, and the present disclosure does not limit its embodiments.

[0034]

[0045] FIG. 2A shows a schematic diagram of an exemplary encoding process 200A according to an embodiment of the present disclosure. For example, the encoding process 200A can be executed by an encoder. As shown in FIG. 2A, the encoder can encode a video sequence 202 into a video bitstream 228 according to process 200A. Similar to the video sequence 100 in FIG. 1, the video sequence 202 can include a set of pictures (referred to as "original pictures") arranged in chronological order. Similar to the structure 110 in FIG. 1, each original picture of the video sequence 202 can be divided into basic processing units, basic processing subunits, or regions for processing by the encoder. In some embodiments, the encoder can execute process 200A at the level of basic processing units for each original picture of the video sequence 202. For example, the encoder can execute process 200A in an iterative manner, and the encoder can encode one basic processing unit in one iteration of process 200A. In some embodiments, the encoder can execute process 200A in parallel for regions (e.g., regions 114-118) of each original picture of the video sequence 202.

[0035]

[0046] In FIG. 2A, the coder can supply the basic processing unit of the original picture of video sequence 202 (referred to as "original BPU") to prediction stage 204 and generate prediction data 206 and prediction BPU 208. The coder can subtract prediction BPU 208 from the original BPU to generate residual BPU 210. The coder can supply residual BPU 210 to transform stage 212 and quantization stage 214 and generate quantized transform coefficients 216. The coder can supply prediction data 206 and quantized transform coefficients 216 to binary coding stage 226 and generate video bitstream 228. Components 202, 204, 206, 208, 210, 212, 214, 216, 226 and 228 can be referred to as the "forward path". During process 200A, after quantization stage 214, the coder can supply quantized transform coefficients 216 to inverse quantization stage 218 and inverse transform stage 220 and generate reconstructed residual BPU 222. The coder can add reconstructed residual BPU 222 to prediction BPU 208 and generate prediction criterion 224 used in prediction stage 204 for the next iteration of process 200A. Components 218, 220, 222 and 224 of process 200A can be referred to as the "reconstruction path". The reconstruction path can be used to ensure that both the coder and the decoder use the same reference data for prediction.

[0036]

[0047] The coder can repeatedly execute process 200A to encode each original BPU of the original picture (within the forward path) and generate prediction criterion 224 for encoding the next original BPU of the original picture (within the reconstruction path). After encoding all the original BPUs of the original picture, the coder can proceed to encode the next picture within video sequence 202.

[0037]

[0048] Referring to process 200A, the coder can receive video sequence 202 generated by a video capture device (e.g., a camera). As used herein, the term "receive" can refer to receiving, inputting, acquiring, obtaining, getting, reading, accessing, or any action by any method for inputting data.

[0038]

[0049] In the prediction stage 204, in the current iteration, the coder can receive the original BPU and prediction criterion 224, perform a prediction operation, and generate prediction data 206 and prediction BPU 208. Prediction criterion 224 can be generated from the reconstruction path of a previous iteration of process 200A. The purpose of prediction stage 204 is to reduce information redundancy by extracting prediction data 206 from prediction data 206 and prediction criterion 224, and prediction data 206 can be used to reconstruct the original BPU as prediction BPU 208.

[0039]

[0050] Ideally, prediction BPU 208 can be the same as the original BPU. However, due to non-ideal prediction and reconstruction operations, prediction BPU 208 generally differs slightly from the original BPU. To record such a difference, after generating prediction BPU 208, the coder can subtract it from the original BPU to generate residual BPU 210. For example, the coder can subtract the value of a pixel of prediction BPU 208 (e.g., a grayscale value or an RGB value) from the value of the corresponding pixel of the original BPU. Each pixel of residual BPU 210 can have a residual value as a result of such subtraction between the corresponding pixels of the original BPU and prediction BPU 208. Compared with the original BPU, prediction data 206 and residual BPU 210 can have fewer bits, but they can be used to reconstruct the original BPU without significant quality degradation. Therefore, the original BPU is compressed.

[0040]

[0051] To further compress the residual BPU 210, in the transformation stage 212, the coder can reduce the spatial redundancy of the residual BPU 210 by decomposing it into a set of two-dimensional "basis patterns", and each basis pattern is associated with a "transformation coefficient". The basis patterns can have the same size (e.g., the size of the residual BPU 210). Each basis pattern can represent a frequency component of the change of the residual BPU 210 (e.g., the frequency of luminance change). None of the basis patterns can be reproduced from any combination (e.g., linear combination) of any other basis patterns. In other words, the decomposition can decompose the change of the residual BPU 210 into the frequency domain. Such a decomposition is similar to the discrete Fourier transform of a function. In this case, the basis patterns are similar to the basis functions of the discrete Fourier transform (e.g., trigonometric functions), and the transformation coefficients are similar to the coefficients associated with the basis functions.

[0041]

[0052] Different transformation algorithms can use different basis patterns. For example, various transformation algorithms such as the discrete cosine transform, the discrete sine transform, or the like can be used in the transformation stage 212. The transformation in the transformation stage 212 is invertible. That is, the coder can recover the residual BPU 210 by the inverse operation of the transformation (referred to as "inverse transformation"). For example, to recover the pixels of the residual BPU 210, the inverse transformation can multiply the corresponding pixel values of the basis patterns by their respective associated coefficients, add the products, and generate a weighted sum. For a video coding standard, both the coder and the decoder can use the same transformation algorithm (and thus the same basis patterns). Therefore, the coder can record only the transformation coefficients, and the decoder can reconstruct the residual BPU 210 from the transformation coefficients without receiving the basis patterns from the coder. Compared with the residual BPU 210, the transformation coefficients can have fewer bits, but they can be used to reconstruct the residual BPU 210 without significant quality degradation. Therefore, the residual BPU 210 is further compressed.

[0042]

[0053] The coder can further compress the transform coefficients at the quantization stage 214. In the transform process, different basis patterns can represent different change frequencies (e.g., luminance change frequency). Since the human eye is generally more adept at recognizing low-frequency changes, the coder can ignore the information of high-frequency changes without causing significant quality degradation during decoding. For example, at the quantization stage 214, the coder can generate the quantized transform coefficients 216 by dividing each transform coefficient by an integer value (referred to as the "quantization scale factor") and rounding the quotient to the nearest integer. After such an operation, some of the transform coefficients of the high-frequency basis pattern can be converted to 0, and the transform coefficients of the low-frequency basis pattern can be converted to smaller integers. The coder can ignore the quantized transform coefficients 216 with a value of 0, thereby further compressing the transform coefficients. The quantization process is also invertible, in which case the quantized transform coefficients 216 can be reconstructed into transform coefficients by the inverse operation of quantization (referred to as "inverse quantization").

[0043]

[0054] Since the coder ignores the remainder of such division in the rounding operation, the quantization stage 214 can be non-invertible. Typically, the quantization stage 214 can contribute to the largest information loss in process 200A. The greater the information loss, the fewer bits are required for the quantized transform coefficients 216. To obtain different information loss levels, the coder can use different values of the quantization parameter or any other parameter of the quantization process.

[0044]

[0055] In the binary encoding stage 226, the coder can encode the prediction data 206 and the quantized transform coefficients 216 using binary encoding techniques such as, for example, entropy encoding, variable-length encoding, arithmetic encoding, Huffman encoding, context-adaptive binary arithmetic encoding, or any other reversible or irreversible compression algorithm. In some embodiments, in addition to the prediction data 206 and the quantized transform coefficients 216, the coder can encode other information in the binary encoding stage 226, such as, for example, the prediction mode used in the prediction stage 204, the parameters of the prediction operation, the type of transformation in the transformation stage 212, the parameters of the quantization process (e.g., quantization parameters), the coder control parameters (e.g., bitrate control parameters), or the like. The coder can generate a video bitstream 228 using the output data of the binary encoding stage 226. In some embodiments, the video bitstream 228 can be further packetized for network transmission.

[0045]

[0056] Referring to the reconstruction path of process 200A, in the inverse quantization stage 218, the coder can perform inverse quantization on the quantized transform coefficients 216 to generate reconstructed transform coefficients. In the inverse transformation stage 220, the coder can generate a reconstructed residual BPU 222 based on the reconstructed transform coefficients. The coder can add the reconstructed residual BPU 222 to the prediction BPU 208 to generate a prediction reference 224 that will be used in the next iteration of process 200A.

[0046]

[0057] Note that other variations of process 200A can be used to encode video sequence 202. In some embodiments, the steps of process 200A can be performed in a different order by the encoder. In some embodiments, one or more steps of process 200A can be combined into a single step. In some embodiments, a single step of process 200A can be divided into multiple steps. For example, the transform step 212 and the quantization step 214 can be combined into a single step. In some embodiments, process 200A can include additional steps. In some embodiments, process 200A can omit one or more steps in FIG. 2A.

[0047]

[0058] FIG. 2B shows a schematic diagram of another exemplary encoding process 200B according to an embodiment of the present disclosure. Process 200B can be changed from process 200A. For example, process 200B can be used by an encoder compliant with a hybrid video encoding standard (e.g., the H.26x series). Compared with process 200A, the forward path of process 200B additionally includes a mode decision step 230 and divides the prediction step 204 into a spatial prediction step 2042 and a temporal prediction step 2044. The reconstruction path of process 200B additionally includes a loop filter step 232 and a buffer 234.

[0048]

[0059] Generally, prediction techniques can be classified into two types: spatial prediction and temporal prediction. Spatial prediction (e.g., intra-picture prediction or "intra prediction") can use pixels from one or more already-encoded adjacent BPUs within the same picture to predict the current BPU. That is, the prediction reference 224 in spatial prediction can include adjacent BPUs. Spatial prediction can reduce the inherent spatial redundancy of a picture. Temporal prediction (e.g., inter-picture prediction or "inter prediction") can use regions from one or more already-encoded pictures to predict the current BPU. That is, the prediction reference 224 in temporal prediction can include encoded pictures. Temporal prediction can reduce the inherent temporal redundancy of a picture.

[0049]

[0060] Referring to process 200B, within the forward path, the encoder performs prediction operations in the spatial prediction stage 2042 and the temporal prediction stage 2044. For example, in the spatial prediction stage 2042, the encoder can perform intra prediction. For the original BPU of the picture being encoded, the prediction reference 224 can include one or more adjacent BPUs that are encoded (within the forward path) and reconstructed (within the reconstruction path) within the same picture. The encoder can generate the predicted BPU 208 by extrapolating the adjacent BPUs. The extrapolation technique can include, for example, linear extrapolation or interpolation, polynomial extrapolation or interpolation, or the like. In some embodiments, the encoder can perform extrapolation at the pixel level, such as by extrapolating the value of each pixel of the predicted BPU 208. The adjacent BPUs used for extrapolation can be located relative to the original BPU in various directions, such as the vertical direction (e.g., above the original BPU), the horizontal direction (e.g., to the left of the original BPU), the diagonal direction (e.g., bottom-left, bottom-right, top-left, or top-right of the original BPU), or any direction defined by the video encoding standard being used. For intra prediction, the prediction data 206 can include, for example, the location (e.g., coordinates) of the adjacent BPUs used, the size of the adjacent BPUs used, the parameters of the extrapolation, the direction of the adjacent BPUs used relative to the original BPU, or the like.

[0050]

[0061] As another example, in the temporal prediction stage 2044, the coder can perform inter prediction. For the current picture's original BPU, the prediction reference 224 can include one or more pictures (referred to as "reference pictures") that are encoded (within the forward path) and reconstructed (within the reconstruction path). In some embodiments, the reference pictures can be encoded and reconstructed for each BPU. For example, the coder can add the reconstructed residual BPU 222 to the prediction BPU 208 to generate a reconstructed BPU. When all the reconstructed BPUs of the same picture have been generated, the coder can generate the reconstructed picture as a reference picture. The coder can perform an operation of "motion estimation" to search for a matching region within a range of reference pictures (referred to as the "search window"). The location of the search window within the reference picture can be determined based on the location of the original BPU of the current picture. For example, the search window can be centered at a location within the reference picture that has the same coordinates as the original BPU within the current picture and can be extended outward over a predetermined distance. When the coder identifies a region similar to the original BPU within the search window (e.g., by using a pixel recursive algorithm, a block matching algorithm, or the like), the coder can determine such a region as the matching region. The matching region can have dimensions that are different from the original BPU (e.g., smaller than, equal to, larger than, or of a different shape than the original BPU). Since the reference picture and the current picture are temporally separated within the timeline (e.g., as shown in FIG. 1), as time passes, the matching region can be considered to "move" to the location of the original BPU. The coder can record such a direction and distance of motion as a "motion vector". When multiple reference pictures are used (e.g., like picture 106 in FIG. 1), the coder can search for a matching region for each reference picture and determine its associated motion vector. In some embodiments, the coder can assign weights to the pixel values of the matching region of each matching reference picture.

[0051]

[0062] Motion estimation can be used to identify various types of motion, such as translation, rotation, zooming, or the like. For inter prediction, the prediction data 206 can include, for example, the location (e.g., coordinates) of the matching region, the motion vector associated with the matching region, the number of reference pictures, the weight associated with the reference pictures, or the like.

[0052]

[0063] To generate the prediction BPU 208, the coder can perform the operation of "motion compensation". Motion compensation can be used to reconstruct the prediction BPU 208 based on the prediction data 206 (e.g., motion vector) and the prediction reference 224. For example, the coder can move the matching region of the reference picture according to the motion vector, whereby the coder can predict the original BPU of the current picture. When multiple reference pictures are used (e.g., picture 106 in FIG. 1), the coder can move the matching regions of the reference pictures according to their respective motion vectors and average the pixel values of the matching regions. In some embodiments, when the coder assigns weights to the pixel values of the matching regions of each matching reference picture, the coder can add the weighted sum of the pixel values to the moved matching region.

[0053]

[0064] In some embodiments, inter prediction can be unidirectional or bidirectional. Unidirectional inter prediction can use one or more reference pictures in the same temporal direction with respect to the current picture. For example, picture 104 in FIG. 1 is a unidirectional inter prediction picture in which the reference picture (i.e., picture 102) precedes picture 104. Bidirectional inter prediction can use one or more reference pictures in both temporal directions with respect to the current picture. For example, picture 106 in FIG. 1 is a bidirectional inter prediction picture in which the reference pictures (i.e., pictures 104 and 108) are in both temporal directions with respect to picture 104.

[0054]

[0065] Referring still to the forward path of process 200B, after the spatial prediction 2042 and the temporal prediction stage 2044, at the mode decision stage 230, the coder can select a prediction mode (e.g., one of intra prediction or inter prediction) for the current iteration of process 200B. For example, the coder can perform a rate-distortion optimization technique. In this technique, the coder can select a prediction mode to minimize the value of a cost function that depends on the bitrate of the candidate prediction mode and the distortion of the reconstructed reference picture under the candidate prediction mode. Depending on the selected prediction mode, the coder can generate the corresponding prediction BPU 208 and prediction data 206.

[0055]

[0066] Within the reconstruction path of process 200B, if the intra prediction mode is selected within the forward path, after generating the prediction reference 224 (e.g., the current BPU encoded and reconstructed in the current picture), the coder can directly supply the prediction reference 224 to the spatial prediction stage 2042 for later use (e.g., for extrapolation of the next BPU of the current picture). The coder can supply the prediction reference 224 to the loop filter stage 232, where the coder can apply a loop filter to the prediction reference 224 to reduce or eliminate the distortion (e.g., blocking artifacts) introduced during the encoding of the inter prediction reference 224. The coder can apply various loop filter techniques, such as deblocking, sample adaptive offset, adaptive loop filter, or the like, at the loop filter stage 232. The loop-filtered reference picture can be stored in the buffer 234 (or "decoded picture buffer") for later use (e.g., for use as an inter prediction reference picture for future pictures of the video sequence 202). The coder can store one or more reference pictures in the buffer 234 for use in the temporal prediction stage 2044. In some embodiments, the coder can encode the parameters of the loop filter (e.g., loop filter strength) at the binary encoding stage 226 together with the quantized transform coefficients 216, the prediction data 206, and other information.

[0056]

[0067] Figure 3A shows a schematic diagram of an exemplary decoding process 300A according to an embodiment of the present disclosure. Process 300A can be a restoration process corresponding to the compression process 200A in FIG. 2A. In some embodiments, process 300A can be similar to the reconstruction path of process 200A. The decoder can decode the video bitstream 228 into a video stream 304 according to process 300A. The video stream 304 can be very similar to the video sequence 202. However, due to information loss in the compression and restoration processes (e.g., the quantization stage 214 in FIGS. 2A-2B), generally, the video stream 304 is not identical to the video sequence 202. Similar to processes 200A and 200B in FIGS. 2A-2B, the decoder can execute process 300A at the level of the basic processing unit (BPU) for each picture encoded in the video bitstream 228. For example, the decoder can execute process 300A in an iterative manner, and the decoder can decode one basic processing unit in one iteration of process 300A. In some embodiments, the decoder can execute process 300A in parallel for each region (e.g., regions 114-118) of each picture encoded in the video bitstream 228.

[0057]

[0068] In FIG. 3A, the decoder can supply a portion of the video bitstream 228 associated with the basic processing unit of the encoded picture (referred to as the "encoded BPU") to the binary decoding stage 302. In the binary decoding stage 302, the decoder can decode such a portion into prediction data 206 and quantization transform coefficients 216. The decoder supplies the quantization transform coefficients 216 to the inverse quantization stage 218 and the inverse transform stage 220, and can generate the reconstructed residual BPU 222. The decoder supplies the prediction data 206 to the prediction stage 204 and can generate the prediction BPU 208. The decoder can add the reconstructed residual BPU 222 to the prediction BPU 208 and generate the prediction reference 224. In some embodiments, the prediction reference 224 can be stored in a buffer (e.g., a decoded picture buffer in computer memory). The decoder can supply the prediction reference 224 to the prediction stage 204 to perform the prediction operation in the next iteration of process 300A.

[0058]

[0069] The decoder can repeatedly execute process 300A to decode each encoded BPU of the encoded picture and generate the prediction reference 224 for encoding the next encoded BPU of the encoded picture. After decoding all the encoded BPUs of the encoded picture, the decoder can output the picture to the video stream 304 for display and proceed to decode the next encoded picture in the video bitstream 228.

[0059]

[0070] In the binary decoding stage 302, the decoder can perform the inverse operation of the binary encoding technique (e.g., entropy encoding, variable-length encoding, arithmetic encoding, Huffman encoding, context-adaptive binary arithmetic encoding, or any other reversible compression algorithm) used by the encoder. In some embodiments, in addition to the predicted data 206 and the quantized transform coefficients 216, the decoder can decode other information, such as, for example, the prediction mode, the parameters of the prediction operation, the type of transform, the parameters of the quantization process (e.g., quantization parameters), the encoder control parameters (e.g., bitrate control parameters), or the like, in the binary decoding stage 302. In some embodiments, when the video bitstream 228 is transmitted in the form of packets through a network, the decoder can depacketize the video bitstream 228 before supplying it to the binary decoding stage 302.

[0060]

[0071] FIG. 3B shows a schematic diagram of another exemplary decoding process 300B according to an embodiment of the present disclosure. The process 300B can be changed from the process 300A. For example, the process 300B can be used by a decoder compliant with a hybrid video encoding standard (e.g., the H.26x series). Compared with the process 300A, the process 300B additionally divides the prediction stage 204 into a spatial prediction stage 2042 and a temporal prediction stage 2044, and additionally includes a loop filter stage 232 and a buffer 234.

[0061]

[0072] In process 300B, for the encoding basic processing unit (referred to as the "current BPU") of the encoded picture being decoded (referred to as the "current picture"), the prediction data 206 decoded from the binary decoding stage 302 by the decoder can include various types of data depending on which prediction mode was used by the encoder to encode the current BPU. For example, if intra prediction was used by the encoder to encode the current BPU, the prediction data 206 can include an intra prediction, parameters of the intra prediction operation, or a prediction mode indicator (e.g., a flag value) indicating the like. The parameters of the intra prediction operation can include, for example, the location (e.g., coordinates) of one or more adjacent BPUs used as references, the size of the adjacent BPUs, extrapolation parameters, the direction of the adjacent BPUs relative to the original BPU, or the like. As another example, if inter prediction was used by the encoder to encode the current BPU, the prediction data 206 can include an inter prediction, parameters of the inter prediction operation, or a prediction mode indicator (e.g., a flag value) indicating the like. The parameters of the inter prediction operation can include, for example, the number of reference pictures associated with the current BPU, the weights respectively associated with the reference pictures, the location (e.g., coordinates) of one or more matching regions in each reference picture, one or more motion vectors respectively associated with the matching regions, or the like.

[0062]

[0073] Based on the prediction mode indicator, the decoder can determine whether to perform spatial prediction (e.g., intra prediction) in the spatial prediction stage 2042 or temporal prediction (e.g., inter prediction) in the temporal prediction stage 2044. Details of performing such spatial or temporal prediction are described in FIG. 2B and will not be repeated here. After performing such spatial or temporal prediction, the decoder can generate the predicted BPU 208. As described in FIG. 3A, the decoder can add the predicted BPU 208 and the reconstructed residual BPU 222 to generate the prediction reference 224.

[0063]

[0074] In process 300B, the decoder can supply the prediction criterion 224 to the spatial prediction stage 2042 or the temporal prediction stage 2044 for performing the prediction operation in the next iteration of process 300B. For example, when the current BPU is decoded using intra prediction in the spatial prediction stage 2042, after generating the prediction criterion 224 (e.g., the decoded current BPU), the decoder can directly supply the prediction criterion 224 to the spatial prediction stage 2042 for later use (e.g., for extrapolation of the next BPU of the current picture). When the current BPU is decoded using inter prediction in the temporal prediction stage 2044, after generating the prediction criterion 224 (e.g., the reference pictures for which all BPUs are decoded), the encoder can supply the prediction criterion 224 to the loop filter stage 232 to reduce or eliminate distortion (e.g., blocking artifacts). The decoder can apply the loop filter to the prediction criterion 224 in the manner described in FIG. 2B. The loop-filtered reference picture can be stored in buffer 234 (e.g., a decoded picture buffer in computer memory) for later use (e.g., for use as an inter prediction reference picture for future encoded pictures of video bitstream 228). The decoder can store one or more reference pictures in buffer 234 for use in the temporal prediction stage 2044. In some embodiments, the prediction data may further include loop filter parameters (e.g., the strength of the loop filter). In some embodiments, if the prediction mode indicator of the prediction data 206 indicates that inter prediction has been used to encode the current BPU, the prediction data includes loop filter parameters.

[0064]

[0075] FIG. 4 is a block diagram of an exemplary apparatus 400 for encoding or decoding video according to an embodiment of the present disclosure. As shown in FIG. 4, apparatus 400 may include a processor 402. When processor 402 executes the instructions described herein, apparatus 400 can become a dedicated machine for video encoding or decoding. Processor 402 can be any type of circuitry having the ability to manipulate or process information. For example, processor 402 can include a central processing unit (or "CPU"), a graphics processing unit (or "GPU"), a neural processing unit ("NPU"), a microcontroller unit ("MCU"), an optical processor, a programmable logic controller, a microcontroller, a microprocessor, a digital signal processor, an intellectual property (IP) core, a programmable logic array (PLA), a programmable array logic (PAL), a generic array logic (GAL), a complex programmable logic device (CPLD), a field programmable gate array (FPGA), a system on chip (SoC), an application specific integrated circuit (ASIC), or any number of any combination of the like. In some embodiments, processor 402 can also be a set of processors grouped as a single logical component. For example, as shown in FIG. 4, processor 402 can include a plurality of processors including processor 402a, processor 402b, and processor 402n.

[0065]

[0076] Device 400 may also include a memory 404 configured to store data (e.g., a set of instructions, computer code, intermediate data, or the like). For example, as shown in FIG. 4, the stored data may include program instructions (e.g., program instructions for implementing the steps in processes 200A, 200B, 300A, or 300B) and data for processing (e.g., video sequence 202, video bitstream 228, or video stream 304). The processor 402 can access program instructions and data for processing (e.g., via bus 410), execute the program instructions, and perform operations or manipulations on the data for processing. The memory 404 may include a high-speed random access storage device or a non-volatile storage device. In some embodiments, the memory 404 may include any number of any combination of random access memory (RAM), read-only memory (ROM), optical disk, magnetic disk, hard drive, solid state drive, flash drive, secure digital (SD) card, memory stick, compact flash (registered trademark) (CF) card, or the like. The memory 404 may also be a group of memories grouped as a single logical component (not shown in FIG. 4).

[0066]

[0077] The bus 410 can be a communication device that transfers data between internal components of the device 400, such as an internal bus (e.g., a CPU-memory bus), an external bus (e.g., a universal serial bus port, a peripheral component interconnect express port), or the like.

[0067]

[0078] For ease of explanation without causing ambiguity, the processor 402 and other data processing circuits are collectively referred to herein as "data processing circuits" in the present disclosure. The data processing circuits can be implemented entirely in hardware or as a combination of software, hardware, and firmware. In addition, the data processing circuits can be a single independent module or can be fully or partially combined with any other component of the device 400.

[0068]

[0079] Device 400 may further include a network interface 406 for providing wired or wireless communication with a network (e.g., the Internet, an intranet, a local area network, a mobile communication network, or the like). In some embodiments, network interface 406 may include any number of any combination of a network interface controller (NIC), a radio frequency (RF) module, a transponder, a transceiver, a modem, a router, a gateway, a wired network adapter, a wireless network adapter, a Bluetooth adapter, an infrared adapter, a near field communication (“NFC”) adapter, a cellular network chip, or the like.

[0069]

[0080] In some embodiments, optionally, device 400 may further include a peripheral interface 408 for providing connection to one or more peripheral devices. As shown in FIG. 4, peripheral devices may include, but are not limited to, a cursor control device (e.g., a mouse, a touchpad, or a touch screen), a keyboard, a display (e.g., a cathode ray tube display, a liquid crystal display, or a light emitting diode display), a video input device (e.g., a camera, or an input interface coupled to a video archive), or the like.

[0070]

[0081] Note that the video codec (e.g., the codec that executes processes 200A, 200B, 300A, or 300B) can be implemented as any combination of any software or hardware modules within device 400. For example, some or all stages of processes 200A, 200B, 300A, or 300B can be implemented as one or more software modules of device 400, such as program instructions loaded into memory 404. As another example, some or all stages of processes 200A, 200B, 300A, or 300B can be implemented as one or more hardware modules of device 400, such as special data processing circuitry (e.g., FPGA, ASIC, NPU, or the like).

[0071]

[0082] The present disclosure provides video encoding and decoding methods that use a cross-component linear model (CCLM), a cross-component non-linear model (CCNLM), or a gradient model to predict chroma samples based on concatenated luma samples. The disclosed models can be freely combined to perform chroma sample prediction. The disclosed cross-component prediction methods can reduce cross-component redundancy and can be used in conjunction with any image / video encoding standard such as Advanced Video Coding (AVC), High Efficiency Video Coding (HEVC), Versatile Video Coding (VVC), AOMedia Video 1 (AV1), Joint Photographic Experts Group (JPEG), Moving Picture Experts Group (MPEG), etc.

[0072]

[0083] According to some embodiments, the CCLM can be used to predict the chroma samples of a block from concatenated reconstructed luma samples by a linear model as in Equation 1.

Equation

Equation

[0073]

[0084] Three CCLM modes (CCLM_LT, CCLM_L, and CCLM_T) are described in detail below. These three modes differ with respect to the location of the reconstructed adjacent samples used for linear model parameter (α and β) derivation. The reconstructed sample adjacent above is involved in the CCLM_T mode, and the reconstructed sample adjacent to the left is involved in the CCLM_L mode. In the CCLM_LT mode, both the reconstructed sample adjacent above and the reconstructed sample adjacent to the left are used.

[0074]

[0085] In the signaling of the chroma intra mode, a flag indicating whether CCLM is applied is signaled first. If the flag is signaled as true, which CCLM mode of the three is applied is signaled further.

[0075]

[0086] In the disclosed embodiments, downsampling of the reconstructed luma samples may be used. To align the positions of the chroma samples in a 4:2:0 or 4:2:2 color format video sequence, two types of downsampling filters as shown in Equations 2 and 3 may be applied to the luma samples, both having a 2-to-1 downsampling ratio in both the horizontal and vertical directions. Based on the SPS level flag, a two-dimensional 6-tap or 5-tap filter is applied not only to the luma samples within the current block but also to its adjacent luma samples. When the SPS level flag is equal to 1, the prediction process operates in a manner designed for the position of the chroma sample that is not shifted vertically relative to the position of the corresponding luma sample, signaling that a 5-tap filter is used. When the SPS level flag is equal to 0, the prediction process operates in a manner designed for the position of the chroma sample that is shifted downward by 0.5 units of the luma sample relative to the position of the corresponding luma sample, signaling that a 6-tap filter is used. An exception occurs when the top row of the current block is the CTU boundary. In this case, a one-dimensional 3-tap filter as shown in Equation 4 is applied to the adjacent luma samples above to avoid using two or more luma rows across the CTU boundary.

Number

[0076]

[0087] The process of downsampling using the aforementioned filters may be represented by the following equations, where Equations 5, 6, and 7 correspond to the filters in Equations 2, 3, and 4 respectively.

Number

Number

[0077]

[0088] The linear model coefficients α and β are derived based on the reconstructed adjacent chroma samples that are downsampled and the corresponding reconstructed luma samples on both the encoder side and the decoder side for non-4:4:4 color formats in order to avoid any signaling overhead.

[0078]

[0089] In the initially adopted version of the CCLM mode, a linear minimum mean square error (LMMSE) estimator was used for parameter derivation.

Equation

Equation

[0079]

[0090] In some embodiments, the number of adjacent samples used to derive the linear model parameters can be increased or decreased in order to ensure that the number of samples used to derive the model parameters is a power of 2.

[0080]

[0091] For example, in order to reduce the complexity of the computer, only four adjacent samples may be used to derive the model parameters. FIG. 6 is a schematic diagram showing a method of using adjacent samples to derive model parameters according to some embodiments of the present disclosure. In FIG. 6, the adjacent samples used to derive the model parameters are shown as circles. As shown in FIG. 6A, for a W×H chroma block (e.g., an 8×8 chroma CU), the four adjacent samples used in the CCLM_LT mode are samples at positions W / 4 and 3W / 4 at the top boundary, and at positions H / 4 and 3H / 4 at the left boundary. As shown in FIG. 6B, for the CCLM_L mode, the left boundary is extended to a (W+H) sample size, and the four samples used for model parameter derivation are at positions (W+H) / 8, 3(W+H) / 8, 5(W+H) / 8, and 7(W+H) / 8. As shown in FIG. 6C, for the CCLM_T mode, the top boundary is extended to a (W+H) sample size, and the four samples used for model parameter derivation are at positions (W+H) / 8, 3(W+H) / 8, 5(W+H) / 8, and 7(W+H) / 8.

[0081]

[0092] The four reconstructed and downsampled adjacent luma samples at the selected positions are two smaller values:

number

number

number

Number

[0082]

[0093] Finally, the linear model coefficients α and β are obtained according to the following equations.

Number

[0083]

[0094] The division operation for calculating the parameter α is performed by a lookup table. To reduce the memory required to store the table, the difference value (the difference between the maximum value and the minimum value) and the parameter α are expressed in exponential notation. For example, the difference is approximated by a 4-bit significant part and an exponent. As a result, the table of 1 / difference is reduced to 16 elements of 16 values of the mantissa as follows. DivTable[]={0,7,6,5,5,4,4,3,3,2,2,1,1,1,1,0} (Equation 13)

[0084]

[0095] This will have both the benefits of reducing not only the computational complexity but also the memory size required to store the required table.

[0085]

[0096] In the disclosed embodiments, a multi-model CCLM can be used. The CCLM can be extended by adding three multi-model CCLM (MMLM) modes (MMLM_LT, MMLM_L, and MMLM_T). The differences between the three modes are the same as the differences between the CCLM_LT, CCLM_L, and CCLM_T modes (i.e., the locations of the reconstructed adjacent samples used for deriving the linear model parameters (α and β)). In each MMLM mode, there can be two or more linear models between the luma and chroma within a block. First, the reconstructed adjacent samples are classified into two classes by using a threshold that is the average of the values of the luma reconstruction adjacent samples. Next, each class is treated as an independent training set for deriving a linear model by using the aforementioned LMMSE method. Thereafter, the reconstructed luma samples of the current block are also classified based on the same rule. Finally, the chroma samples are predicted by the reconstructed luma samples in different ways for different classes.

[0086]

[0097] In the disclosed embodiments, a gradient linear model (GLM) method can be used. Compared with CCLM, instead of the downsampling reconstructed luma samples, the GLM utilizes the luma sample gradient for deriving a linear model. In other words, instead of using filters in Equations 4 - 6, the gradient G is used in the CCLM process. Other designs of CCLM (e.g., parameter derivation, prediction sample linear transformation) are maintained without change. The gradient G can be calculated by one of four Sobel-based gradient patterns.

Number

[0087]

[0098] By using the aforementioned gradient patterns, the gradient G can be calculated by the following equations, where Equations 18, 19, 20, and 21 respectively correspond to the gradient patterns in Equations 14, 15, 16, and 17.

Number

[0088]

[0099] The linear model parameters α and β are derived based on the reconstructed adjacent chroma samples and the corresponding gradient G of the concatenated reconstructed luma samples on both the encoder side and the decoder side by the same method of CCLM (e.g., the LMMSE method). Next, the chroma samples of the block can be predicted from the gradient of the concatenated reconstructed luma samples by the following linear model. pred c (i,j)=α·G l (i,j)+β (Equation 22)

[0089]

[0100] Regarding signaling, when the CCLM mode is enabled for the current CU, two flags are signaled separately for the Cb and Cr components to indicate whether the GLM is enabled for the component. When the GLM is enabled for one component, one syntax element is further signaled to select one of four gradient patterns for gradient calculation. In some embodiments, a 2-bit fixed-length code is used to encode the syntax element.

[0090]

[0101] In some embodiments, the GLM is available only for some of the CCLM modes. For example, the GLM is available only for the CCLM_LT mode, i.e., for the CCLM_LT mode, some syntax elements are signaled to indicate whether the GLM is enabled and which gradient pattern is used. When the GLM is enabled for the CCLM_LT mode, the gradients G of the reconstructed lumasamples adjacent above and the reconstructed lumasamples adjacent to the left are used to replace the adjacent downsampled reconstructed lumasamples in the linear model parameter derivation process, and only the signal linear model is used in the current block. When the GLM is disabled for the CCLM_LT mode, the original CCLM_LT mode is applied. There is no change for the other CCLM modes (i.e., CCLM_L, CCLM_T, and the three MMLM modes). For another example, the GLM is available only for the CCLM_LT mode and the MMLM_LT mode. For another example, the GLM is available only for the CCLM_LT mode, the CCLM_L mode, and the CCLM_T mode. For another example, the GLM is available for all of the six CCLM modes.

[0091]

[0102] When the GLM is applied to the MMLM mode, the multi-model GLM (MMGLM) method is used. In the MMGLM mode, there may be two or more linear models between the gradient G and the chromasamples in the block. When implementing the MMGLM method, the gradients of the reconstructed adjacent samples are first classified into two classes by using a threshold that is the average of the values of the gradients of the luma-reconstructed adjacent samples. Next, each class is treated as an independent training set for deriving a linear model by using the LMMSE method described above. Then, the gradients of the reconstructed lumasamples of the current block are also classified based on the same rule. Finally, the chromasamples are predicted by the gradients of the reconstructed lumasamples in different ways for different classes.

[0092]

[0103] In some embodiments, 16 gradient patterns are supported for the GLM method. That is, the gradient G can be calculated by one of the 16 gradient patterns according to the following formula. The syntax element is signaled to indicate which of the 16 gradient patterns is used.

Number

[0093]

[0104] In some embodiments, the downsampled reconstructed lumasamples and the gradient of the reconstructed lumasamples are used together to derive a linear model. The linear model parameters α and β are derived based on the reconstructed adjacent chromasamples on both the encoder side and the decoder side, the corresponding gradient G of the concatenated reconstructed lumasamples, and the downsampled reconstructed lumasamples by the same method as CCLM (e.g., the LMMSE method). The value of the downsampled reconstructed lumasamples can be obtained by one of the downsampled filters described above. Next, the chromasamples of the block can be predicted by a linear model as follows from the gradient of the concatenated reconstructed lumasamples and the value of the downsampled reconstructed lumasamples.

Number

[0094]

[0105] In some embodiments, one flag is signaled to indicate which of the GLM methods in Formulas 23-38 is used.

[0095]

[0106] In research on video coding techniques, it has been recognized that there is a correlation between various color components. CCLM assumes that there is a linear correlation between chroma samples and luma samples at corresponding positions within an encoding block. However, this linear relationship may not be suitable for all encoding blocks. Sometimes, even for local textures, there can be a more complex relationship between luma and chroma, and in this case, it may be more preferable to use a non-linear relationship for the assumption.

[0096]

[0107] Furthermore, in the GLM method, chroma samples are predicted by constructing a relationship between the gradient of the reconstructed luma samples and the gradient of the chroma samples. However, this method can make it difficult to accurately predict chroma samples when the gradient of the reconstructed luma samples of the current block is very close. Even if the values of the downsampled reconstructed luma samples are introduced into the GLM method corresponding to Equation 39, these values always have the same parameters with the gradient of the reconstructed luma samples, which is not suitable in some cases.

[0097]

[0108] In the present disclosure, it is proposed to predict the chroma samples of a block from the concatenated reconstructed luma samples by using a non-linear model.

[0098]

[0109] In an exemplary embodiment, the Cross-Component Non-Linear Model (CCNLM) is used to predict the chroma samples of the current block from the downsampled reconstructed luma samples, as shown in Equation 40.

Number

Number

[0099]

[0110] According to some embodiments, some terms in Equation 40 can be removed. For example, if the value of n is equal to 2 and the terms with a power of 1 are removed, Equation 40 can be rewritten as follows.

Number

[0100]

[0111] According to some embodiments, in order to make the coefficients of each term of the same order of magnitude, each term can be multiplied by different adjustment coefficients as in the following equation.

Number

[0101]

[0112] According to some embodiments, some coefficients in Equation 40 are restricted to be the same. For example, if the value of n is equal to 2 and the terms with a power of 1 and the terms with a power of 2 have the same coefficient, Equation 40 can be rewritten as follows.

Number

[0102]

[0113] According to some embodiments, the foregoing embodiments may be freely combined. For example, the value of n is equal to 2, and each term is multiplied by different adjustment coefficients as in Equation 44, where factor0 = bitDepth / 2, factor1 = 1, and factor2 = 1 / bitDepth. For another example, the value of n is equal to 2, and each term is multiplied by different adjustment coefficients. The terms with a power of 1 and the terms with a power of 2 have the same coefficients as in Equation 45, where factor0 = bitDepth / 2, factor1 = 1, and factor2 = 1 / bitDepth.

Number

[0103]

[0114] According to some embodiments, the model parameters a0,..., a n are derived based on the original chroma samples within the current block and the concatenated original luma samples of the same block that are downsampled and signaled in the bitstream for non-4:4:4 color formats in the encoder. In the decoder, the model parameters a0,..., a n are decoded from the bitstream.

[0104]

[0115] According to some embodiments, the model parameters a0,..., a n are derived based on the reconstructed adjacent chroma samples that are downsampled for non-4:4:4 color formats on both the encoder side and the decoder side to avoid any signaling overhead, and their corresponding reconstructed luma samples.

[0105]

[0116] For example, the adjacent samples used to derive the non-linear model parameters in the proposed method may be the same as the adjacent samples used to derive the linear model parameters in the aforementioned CCLM method.

[0106]

[0117] For another example, the adjacent samples used to derive the non-linear model parameters by the proposed method can be the reconstructed samples within the x adjacent rows and columns of the current block, where x can be any positive integer (e.g., x = 3).

[0107]

[0118] For another example, a subset of adjacent samples is used to derive the non-linear model parameters by the proposed method (e.g., four samples).

[0108]

[0119] According to some embodiments, the model parameters a0,..., a n are derived by the Least Mean Squares (LMS) method. The LMS method derives the parameters a0,..., a n by minimizing the Mean Squared Error (MSE) between the predicted and reconstructed values of adjacent chroma samples. Specifically, the initial values of the parameters a0,..., a n can be used to derive the predicted values of adjacent chroma samples. Next, the values of the parameters a0,..., a n can be adjusted by minimizing the MSE between the predicted and reconstructed values of adjacent chroma samples.

[0109]

[0120] For example, the gradient descent method can be used to minimize the MSE. As understood in the art, the gradient descent method is an iterative first-order optimization algorithm that can find the minimum / maximum of a given function (e.g., the MSE between the predicted and reconstructed values of adjacent chroma samples).

[0110]

[0121] For another example, the system of equations for optimizing the partial derivative function of the MSE is described in matrix multiplication form including first-order equations, and the model parameters can be derived by solving the first-order equations by the Gaussian elimination method.

[0111]

[0122] For another example, a self - correlation matrix of the reconstructed values of the down - sampled adjacent luma samples and a cross - correlation vector between the reconstructed values of the down - sampled adjacent luma samples and the reconstructed values of the adjacent chroma samples are calculated. The self - correlation matrix is LU - decomposed, LDL - decomposed, or Cholesky - decomposed, and the parameters a0,...,a n are calculated by using back substitution.

[0112]

[0123] According to some embodiments, similar to the aforementioned CCLM method used to derive adjacent samples' model parameters a0,...,a n , the proposed CCNLM method can support various modes based on the locations of adjacent samples. For example, three CCNLM modes (CCNLM_LT, CCNLM_L, and CCNLM_T) are supported. These three modes are different with respect to the locations of the reconstructed adjacent samples used for deriving the non - linear model parameters (a0,...,a n ). The samples adjacent above the reconstruction are involved in the CCNLM_T mode, and the samples adjacent to the left of the reconstruction are involved in the CCNLM_L mode. In the CCNLM_LT mode, both the samples adjacent above the reconstruction and the samples adjacent to the left of the reconstruction are used.

[0113]

[0124] According to some embodiments, similar to the aforementioned CCLM method, a multi-model CCNLM (MMNLM) method is proposed. Specifically, there may be two or more non-linear models between the luma and chroma within a block. For example, adjacent samples are used to derive model parameters. First, the reconstructed adjacent samples are classified into two classes by using a threshold (e.g., the average of the values of the luma reconstructed adjacent samples). Next, each class is treated as an independent training set for deriving a non-linear model by using the aforementioned method. Then, the reconstructed luma samples of the current block are also classified based on the same rule. Finally, the chroma samples are predicted by the reconstructed luma samples in different ways for different classes.

[0114]

[0125] In some embodiments, MMNLM can support various modes based on the location of adjacent samples when the adjacent samples are used to derive model parameters a0,...,a n . For example, three MMNLM modes (MMNLM_LT, MMNLM_L, and MMNLM_T) are supported. These three modes are different with respect to the location of the reconstructed adjacent samples used for deriving non-linear model parameters (a0,...,a n ). The reconstructed adjacent sample above is involved in the MMNLM_T mode, and the reconstructed adjacent sample to the left is involved in the MMNLM_L mode. In the MMNLM_LT mode, both the reconstructed adjacent sample above and the reconstructed adjacent sample to the left are used.

[0115]

[0126] According to some embodiments, the following six CCLM modes are supported: CCLM_LT, CCLM_L, CCLM_T, MMLM_LT, MMLM_L, and MMLM_T. In the disclosed embodiments, various modes of the proposed CCNLM can be used to replace some or all of the CCLM modes.

[0116]

[0127] According to some embodiments, the proposed CCNLM is used to replace all of the aforementioned CCLM modes. That is, six CCNLM modes (CCNLM_LT, CCNLM_L, CCNLM_T, MMNLM_LT, MMNLM_L, and MMNLM_T) are used to replace the six CCLM modes respectively. Also, there is no additional syntax signaling.

[0117]

[0128] According to some embodiments, the proposed CCNLM is used to replace some of the aforementioned CCLM modes.

[0118]

[0129] For example, the proposed CCNLM is used to replace the CCLM_LT mode. As a result, CCNLM_LT, CCLM_L, CCLM_T, MMLM_LT, MMLM_L, and MMLM_T are supported. A non-linear model between luma and chroma is derived for the CCNLM_LT mode, and a linear model is derived for the other modes.

[0119]

[0130] Regarding another example, the proposed CCNLM is used to replace the CCLM_LT mode and the MMLM_LT mode. As a result, CCNLM_LT, CCLM_L, CCLM_T, MMNLM_LT, MMLM_L, and MMLM_T are supported. A non-linear model between luma and chroma is derived for the CCNLM_LT mode and the MMLM_LT mode, and a linear model is derived for the other modes.

[0120]

[0131] Regarding another example, the proposed CCNLM is used to replace the CCLM_LT mode, the CCLM_L mode, and the CCLM_T mode. As a result, CCNLM_LT, CCNLM_L, CCNLM_T, MMLM_LT, MMLM_L, and MMLM_T are supported. A non-linear model between luma and chroma is derived for the CCNLM_LT mode, the CCNLM_L mode, and the CCNLM_T mode, and a linear model is derived for the other modes.

[0121]

[0132] According to some embodiments, an explicit signaling method is used to determine whether to use the proposed CCNLM method (i.e., whether to use a linear model or a non-linear model). For example, after signaling the CCLM mode, one flag is signaled to indicate whether to use the proposed CCNLM method. In other words, the first few syntax elements are signaled to indicate the position of the adjacent samples used in model parameter derivation (LT, L, or T) and whether to use a single model or a dual model within the current block. Next, one flag is signaled to indicate whether to use a linear model or a non-linear model.

[0122]

[0133] According to some embodiments, only some of the six CCNLM modes are supported by the explicit signaling method.

[0123]

[0134] According to some embodiments, two flags are signaled to indicate whether to use the proposed CCNLM methods for Cb and Cr, respectively.

[0124]

[0135] According to some embodiments, an implicit method is used to determine whether to use the proposed CCNLM method (i.e., whether to use a linear model or a non-linear model).

[0125]

[0136] For example, based on the reconstructed adjacent samples, both a linear model and the proposed non-linear model are derived. Next, the two models are used to predict the adjacent chroma samples from the reconstructed adjacent luma samples. The predicted values obtained by the two types of models are used to calculate the sum of absolute differences (SAD) or the sum of absolute transform differences (SATD) between the reconstructed values of the adjacent chroma samples. Finally, the model with the smaller SAD or SATD is used to predict the chroma samples of the current block.

[0126]

[0137] For another example, the gradients of adjacent luma samples and adjacent chroma samples are used to determine whether a linear model or a non-linear model should be used.

[0127]

[0138] For another example, the size of the current chroma block is used to determine whether a linear model or a non-linear model should be used. For example, if the area of the current chroma block is greater than a threshold (e.g., 256), a non-linear model is used; otherwise, a linear model is used.

[0128]

[0139] In the disclosed embodiments, the CCNLM method can be combined with the GLM method. According to some embodiments, the proposed CCNLM method is combined with the GLM method, and the chroma samples of the current block are predicted from the gradient G of the reconstructed luma samples concatenated as shown in Equation 46. pred c (i,j)=a n ·(G l (i,j)) n +a n-1 ·(G l (i,j)) n-1 +···+a1·G l (i,j)+a0 (Equation 46) Here, pred C (i,j) represents the predicted value of the chroma sample within the current block. Also, G L (i,j) represents the corresponding gradient G of the concatenated reconstructed luma samples of the same block, (i,j) are the coordinates of the samples within the block, and the coefficients a0,...,a n are non-linear model parameters that can represent a non-linear model, and the value of n can be any positive integer greater than 1 (e.g., n = 2).

[0129]

[0140] According to some embodiments, some terms in Equation 46 can be removed. For example, if the value of n is equal to 2 and the term with the power of 1 is removed, Equation 46 can be rewritten as follows. pred c(i,j) = a2·(G l (i,j)) 2 + a0 (Equation 47)

[0130]

[0141] According to some embodiments, in order to make the coefficients of each term in Equation 46 of the same order of magnitude, each term can be multiplied by different adjustment coefficients as follows. pred c (i,j) = a n ·factor n ·(G l (i,j)) n + a n-1 ·factor n-1 ·G l (i,j)) n-1 + ··· + a l ·factor l ·G l (i,j) + a0·factor0 (Equation 48) For example, factor0 = bitDepth / 2, factor1 = 1, factor m = 1 / bitDepth m-1 (where 1 < m <= n).

[0131]

[0142] According to some embodiments, in some terms of Equation 46, the gradient G L (i,j) can be replaced by

Number

Number

[0132]

[0143] For example, only G(i,j) in the topmost term

Number

[0133]

[0144] For another example, only G(i,j) in the least significant term is [Number] is replaced by [Number]

[0134]

[0145] For another example, only G(i,j) in the least significant term is [Number] not replaced by, and for other terms, G(i,j) is [Number] is replaced by [Number]

[0135]

[0146] According to some embodiments, in some terms in Equation 46, the gradient G L (i,j) can be replaced by the reconstructed value rec L (2i,2j) of the concatenated lum samples.

[0136]

[0147] According to some embodiments, the above four embodiments can be freely combined.

[0137]

[0148] For example, n is equal to 2, adjustment coefficients are used, and factor0 = bitDepth / 2, factor1 = 1, factor2 = 1 / bitDepth. pred c (i,j) = a2·(G l (i,j)) 2 / bitDepth + a l ·G l (i,j) + a0·bitDepth / 2 (Equation 52)

[0138]

[0149] For another example, in the above example, G(i,j) in the most significant term is

Number

Number

[0139]

[0150] The samples used to derive the non - linear model can be the original chroma samples within the current block and the concatenated original luma samples of the same block that are downsampled and signaled in the bitstream for non - 4:4:4 color formats in the encoder. Alternatively, the samples used to derive the non - linear model are derived based on the reconstructed adjacent chroma samples and their corresponding luma samples that are downsampled for non - 4:4:4 color formats on both the encoder side and the decoder side to avoid any signaling overhead.

[0140]

[0151] The non - linear model derivation method can be one of the above - mentioned least - mean - square (LMS) methods.

[0141]

[0152] According to some embodiments, the proposed CCNLM method is always combined with the GLM method. That is, when the GLM method is used for chroma blocks, the non - linear model is used to predict chroma samples.

[0142]

[0153] According to some embodiments, an explicit signaling method is used to determine whether to combine the proposed CCNLM method and the GLM method (i.e., whether to use the linear model or the non-linear model of GLM). For example, after signaling the GLM flag, if the GLM mode is enabled, one flag is further signaled to indicate whether to use the non-linear model.

[0143]

[0154] According to some embodiments, an explicit method is used to determine whether to combine the proposed CCNLM method and the GLM method (i.e., whether to use the linear model or the non-linear model of GLM). For example, the explicit method can be the above-mentioned explicit method used to determine whether to use the proposed CCNLM method (i.e., whether to use the linear model or the non-linear model).

[0144]

[0155] According to some embodiments, the proposed CCNLM method can be combined with only some GLM modes. For example, the GLM method is available for the CCLM_LT mode, the CCLM_L mode, and the CCLM_T mode, and the CCNLM can be combined with the GLM only when the CCLM_LT mode is selected. For another example, the GLM method is available for all six CCLM modes, and the CCNLM can be combined with the GLM only when the CCLM_LT mode is selected.

[0145]

[0156] According to some embodiments, the foregoing embodiments can be freely combined.

[0146]

[0157] For example, the GLM is available only for the CCLM_LT mode, and when the GLM is enabled, the proposed CCNLM method is always combined with the GLM method as shown in Equation 52. The LDL decomposition method is used to derive non-linear parameters from the gradients of the reconstructed adjacent chroma samples and the reconstructed adjacent luma samples.

[0147]

[0158] The present disclosure also provides a new GLM method, in which the chroma samples of the current block are predicted from the gradient G of the concatenated reconstructed luma samples and the gradient G of the downsampled reconstructed luma samples. The prediction is based on Equation 54.

Number

Number

Number

[0148]

[0159] In some embodiments, another GLM method as shown in Equation 55 is proposed.

Number

[0149]

[0160] The coefficients in the GLM methods corresponding to Equations 54 and 55 respectively can be derived based on the values of the reconstructed adjacent chroma samples, the corresponding gradient G of the adjacent concatenated reconstructed luma samples, and the values of the corresponding reconstructed adjacent luma samples. This derivation is performed on both the encoder side and the decoder side by one of the aforementioned methods that use the least mean square (LMS) to derive the model parameters. For example, Gaussian elimination or LDL decomposition can be used to derive the coefficients.

[0150]

[0161] In some embodiments, only one of the aforementioned GLM methods is supported.

[0151]

[0162] For example, the GLM method corresponding to Equation 54 is supported. Regarding the chroma block, one flag is signaled within the bitstream to indicate whether GLM is used. If the flag indicates that GLM is used, the chroma samples within the current block are predicted by using the GLM method corresponding to Equation 54.

[0152]

[0163] In some embodiments, two GLM methods are supported, and one flag is signaled to indicate which of the two GLM methods is used.

[0153]

[0164] In one example, the GLM methods corresponding to Equation 22 and Equation 54 are supported. Regarding the chroma block, one flag is signaled within the bitstream to indicate whether GLM is used. If the flag indicates that GLM is used, another flag is signaled to indicate which of the two GLM methods corresponding to Equation 22 and Equation 54, respectively, is used to predict the chroma samples within the current block.

[0154]

[0165] In another example, the GLM methods corresponding to Equation 39 and Equation 54 are supported, respectively.

[0155]

[0166] In another example, the GLM methods corresponding to Equation 54 and Equation 55 are supported, respectively.

[0156]

[0167] In some embodiments, three GLM methods are supported, and the syntax elements are signaled to indicate which of the GLM methods is used.

[0157]

[0168] For example, GLM methods corresponding to Equation 22, Equation 39, and Equation 54 are supported. For a chroma block, one flag is signaled in the bitstream to indicate whether GLM is used. If the flag indicates that GLM is used, syntax elements are signaled to indicate which one of the three GLM methods corresponding to Equation 22, Equation 39, and Equation 54, respectively, is used to predict the chroma samples in the current block.

[0158]

[0169] In some embodiments, four GLM methods are supported, and syntax elements are signaled to indicate which GLM method is used.

[0159]

[0170] For example, GLM methods corresponding to Equation 22, Equation 39, Equation 54, and Equation 55 are supported. For a chroma block, one flag is signaled in the bitstream to indicate whether GLM is used. If the flag indicates that GLM is used, syntax elements are signaled to indicate which one of the three GLM methods corresponding to Equation 22, Equation 39, Equation 54, and Equation 55, respectively, is used to predict the chroma samples in the current block.

[0160]

[0171] In some embodiments, for the two GLM methods corresponding to Equation 54 and Equation 55, respectively, each term may be multiplied by a different adjustment coefficient according to Equation 56 and Equation 57, respectively. For example, factor2 is equal to 1, factor1 is equal to 1, and factor0 is equal to 512.

Number

[0161]

[0172] In some embodiments, the value of the downsampled reconstructed luma sample in the GLM method corresponding to Equation 54

Number

[0162]

[0173] In some embodiments, the value of the downsampled reconstructed lumasample in the GLM method corresponding to Equation 55 [Number] is the n-th power of the value of the downsampled reconstructed lumasample [Number] and can be replaced by

[0163]

[0174] In one example, only the [Number] in the first term is [Number] replaced by [Number]

[0164]

[0175] In another example, only the [Number] in the second term is [Number] is replaced by

Number

[0165]

[0176] In another example, all

Number

Number

Number

[0166]

[0177] In some embodiments, the values of the downsampled reconstruction lumasamples in the two GLM methods corresponding to Equation 54 and Equation 55 respectively

Number

Number

[0167]

[0178] In the GLM method described above, the gradient G can be calculated by one of the four Sobel-based gradient patterns shown in Equations 14 - 17, or one of the 16 gradient patterns shown in Equations 23 - 38.

[0168]

[0179] In some embodiments, another new GLM method is proposed, and the chroma samples of the current block are predicted from multiple gradients of the concatenated reconstructed luma samples, as shown in Equation 62. pred c (i,j)=a n ·G lK (i,j)+···+a1·G l1 (i,j)+a0 (Equation 62) Here, pred C (i,j) represents the predicted value of the chroma sample within the current block, G L1 (i,j)~G Ln (i,j) represents k different gradients of the concatenated reconstructed luma samples of the same block, (i,j) are the coordinates of the samples within the block, and the coefficients a0~a n are model parameters that can be derived based on the values of the reconstructed adjacent chroma samples and the k different gradients of the adjacent concatenated reconstructed luma samples on both the encoder side and the decoder side, by one of the above methods using the least mean square (LMS) to derive the model parameters. For example, the LDL decomposition method can be used to derive the coefficients.

[0169]

[0180] In one example, two gradients are used together to derive the predicted value of the chroma sample, as shown in Equation 63. The two gradients G L1 (i,j) and G L2 (i,j) can be calculated according to the horizontal gradient pattern and the vertical gradient pattern, as shown in Equations 14 and 15. pred c (i,j)=a2·G l2 (i,j)+a1·G l1 (i,j)+a0 (Equation 63)

[0170]

[0181] In another example, four gradients are used together to derive the predicted value of the chroma sample, as shown in Equation 64. The four gradients G L1 (i,j)~G L4 (i,j) can be calculated according to four gradient patterns, as shown in Equations 14~17. pred c (i,j) = a4·G l4 (i,j) + a3·G l3 (i,j) + a2·G l2 (i,j) + a1·G l1 (i,j) + a0 (Equation 64)

[0171]

[0182] In another example, the number of gradients and gradient patterns used together to derive the predicted value of the chroma sample can be determined by the signaled syntax elements.

[0172]

[0183] In some embodiments, another new GLM method is proposed, and the chroma samples of the current block are predicted from a plurality of gradients of the reconstructed luma samples concatenated as shown in Equation 65 and the values of the concatenated downsampled reconstructed luma samples.

Number

Number

[0173]

[0184] In one example, the two gradients of the concatenated reconstructed lumasamples and the values of the concatenated downsampled reconstructed lumasamples are used together to derive the predicted value of the chromasample, as shown in Equation 66. The two gradients G L1 (i,j) and G L2 (i,j) can be calculated according to the horizontal gradient pattern and the vertical gradient pattern, as shown in Equations 14 and 15. [Number]

[0174]

[0185] In another example, the four gradients of the concatenated reconstructed lumasamples and the values of the concatenated downsampled reconstructed lumasamples are used together to derive the predicted value of the chromasample, as shown in Equation 67. The four gradients G L1 (i,j) to G L4 (i,j) can be calculated according to the gradient pattern, as shown in Equations 14 to 17. [Number]

[0175]

[0186] In another example, the number of gradients and gradient patterns used together to derive the predicted value of the chromasample can be determined by the signaled syntax element.

[0176]

[0187] In some embodiments, a non-linear term of the value of the concatenated downsampled reconstructed lumasample is added to calculate the chromasample. For example, as represented by Equation 68, the non-linear term is added to the GLM method corresponding to Equation 62 to calculate the chromasample. In another example, as represented by Equation 69, the non-linear term is added to the GLM method corresponding to Equation 65 to calculate the chromasample. The value of x can be any positive integer greater than 1. In one example, the value of x is equal to 2. [Number]

[0177]

[0188] In some embodiments, both of the two GLM methods corresponding to Equation 62 and Equation 65 are supported, and the block-level flag is signaled in the bitstream to indicate which GLM method is used to predict the current chroma block. That is, the block-level flag is signaled to indicate whether the values of the concatenated downsampled reconstructed luma samples should be used to predict the chroma samples.

[0178]

[0189] In some embodiments, the number of supported gradients is modified.

[0179]

[0190] In one example, only one gradient is supported. For example, only the horizontal gradient pattern corresponding to Equation 14 is supported.

[0180]

[0191] In another example, the slice-level syntax element is signaled to indicate the number of gradient patterns supported within the current slice.

[0181]

[0192] In some embodiments, the encoding method of the gradient pattern is modified. A truncated unary code is used to encode which of the gradient patterns is used. For example, four gradient patterns as shown in Equations 14 - 17 are supported, and a truncated unary code indicating which of the four gradient patterns is used, as shown in Table 1, is used to encode the syntax element.

[0182]

Table 1

[0183]

[0193] In some embodiments, the Cb and Cr components may share the same GLM enable flag.

[0184]

[0194] In one example, one flag is signaled to indicate whether the GLM is used for both the Cb and Cr components. If the flag indicates that the GLM is used, two syntax elements are signaled to indicate which gradient pattern is used for Cb and Cr respectively.

[0185]

[0195] In another example, one flag is signaled to indicate whether the GLM is used for both the Cb and Cr components. If the flag indicates that the GLM is used, one syntax element is signaled to indicate which gradient pattern is used for both Cb and Cr.

[0186]

[0196] In some embodiments, multiple lines are used to derive the GLM model parameters. Specifically, n chroma adjacent lines may be used to derive the GLM model parameters, and the value of n may be equal to any positive integer greater than 1. For example, the value of n is equal to 6.

[0187]

[0197] In one example, the GLM supports the CCLM_LT, CCLM_L, and CCLM_T modes, and six chroma adjacent lines are used to derive the GLM parameters. FIG. 7 is a schematic diagram showing a method of using samples on adjacent lines to derive the parameters of a gradient model according to some embodiments of the present disclosure. FIG. 7 shows the samples on the adjacent lines of the chroma block as circles. Specifically, as shown in FIG. 7, for a W×H chroma block, when the GLM is used for the CCLM_LT mode, the samples of the upper six lines and the left six lines are used to derive the GLM parameters (FIG. 7A); when the GLM is used for the CCLM_L mode, the samples of the left six lines and the lower left six lines are used to derive the GLM parameters (FIG. 7B); when the GLM is used for the CCLM_T mode, the samples of the upper right six lines are used to derive the GLM parameters (FIG. 7C).

[0188]

[0198] In another example, the GLM supports CCLM_LT, CCLM_L, and CCLM_T modes, and six chroma adjacent lines are used to derive the GLM parameters. FIG. 8 is a schematic diagram showing how samples on adjacent lines are used to derive the parameters of the gradient model according to some embodiments of the present disclosure. FIG. 8 shows the samples on the adjacent lines of the chroma block as circles. Specifically, as shown in FIG. 8, for a W×H chroma block, when the GLM is used for the CCLM_LT mode, the samples of the upper six lines, the upper right six lines, the left six lines, and the lower left six lines are used to derive the GLM parameters (FIG. 8A); when the GLM is used for the CCLM_L mode, the samples of the left six lines and the lower left six lines are used to derive the GLM parameters (FIG. 8B); when the GLM is used for the CCLM_T mode, the samples of the upper six lines and the upper right six lines are used to derive the GLM parameters (FIG. 8C).

[0189]

[0199] In some embodiments, when the GLM is used for the CCLM_LT mode, the samples adjacent to the upper left are also included in the samples used to derive the GLM model parameters.

[0190]

[0200] In some embodiments, the aforementioned extension to the number of lines for deriving the GLM model parameters is only applicable to some GLM modes. For example, two GLM methods corresponding to Equation 22 and Equation 54 can be supported respectively, and a flag is signaled to indicate which GLM method is used for the current block. Next, when the GLM method corresponding to Equation 22 is used, only one adjacent line is used to derive the GLM model parameters, and when the GLM method corresponding to Equation 54 is used, six adjacent lines are used to derive the GLM model parameters.

[0191]

[0201] In some embodiments, a variant form of the multi-model GLM method is proposed. In the multi-model GLM method, two or more models can be used to predict chroma samples within a block. Classification can be based on the value of the gradient G, the sign of the gradient G, the absolute value of the gradient G, or the value of the downsampled reconstructed luma samples

Number

Number

[0192]

[0202] In one example, the gradients of the reconstructed adjacent samples are first classified into two classes by the sign of the gradient. That is, the positive gradient is within the first class, and the negative gradient is within the second class. Next, each class is treated as an independent training set for deriving a linear model by using the aforementioned LMMSE method. Thereafter, the gradients of the reconstructed luma samples of the current block are also classified based on the same rule. Finally, the chroma samples are predicted by the gradients of the reconstructed luma samples in different ways for different classes according to the method in Equation 22.

[0193]

[0203] In another example, the gradients of the reconstructed adjacent samples are first classified into two classes by using a threshold that is the average of the absolute values of the gradients of the luma-reconstructed adjacent samples. That is, gradients with an absolute value greater than or equal to the threshold are classified into the first class, and gradients with an absolute value smaller than the threshold are classified into the second class. Next, each class is treated as an independent training set to derive a linear model by using the aforementioned LMMSE method. Thereafter, the gradients of the reconstructed luma samples of the current block are also classified based on the same rule. Finally, for the first class, the chroma samples are predicted from the gradients according to Equation 22, and for the second class, the chroma samples are predicted from the downsampled reconstructed luma samples according to Equation 1.

[0194]

[0204] In some embodiments, the foregoing embodiments related to the disclosed GLM method may be freely combined.

[0195]

[0205] In one example, only one GLM method corresponding to Equation 54 is supported. For a chroma block, one flag is signaled to indicate whether the GLM is used for both the Cb and Cr components. If the flag indicates that the GLM is used for both the Cb component and the Cr component, one syntax element is signaled by a shortened single-term code to indicate which of the four gradient patterns corresponding to Equations 14 to 17 is used for both Cb and Cr. Six adjacent lines are used to derive the model parameters.

[0196]

[0206] In another example, two GLM methods corresponding to Equation 22 and Equation 54 are each supported. For the chroma block, one flag is signaled to indicate whether the GLM is used for both the Cb and Cr components. If the flag indicates that the GLM is used, another flag is signaled to indicate which of the two GLM methods is used, and one syntax element is signaled by a shortened single-term code to indicate which of the four gradient patterns corresponding to Equations 14 - 17 is used for both Cb and Cr. When the GLM method corresponding to Equation 22 is used, only one adjacent line is used to derive the model parameters, and when the GLM method corresponding to Equation 54 is used, six adjacent lines are used to derive the model parameters.

[0197]

[0207] In some embodiments, when predicting a chroma sample, an offset can be subtracted from the terms of the downsampled reconstructed luma sample. The offset can be different for each block. For example, the offset can be equal to the first downsampled reconstructed luma sample within the current block or can be related to the bit depth. For example, the method in Equation 54 can be modified as follows.

Number

[0198]

[0208] The above-described embodiments may be performed as part of a video data processing process such as an encoding process or a decoding process. FIG. 9 is a flowchart of a method 900 for predicting chroma samples by using an inter-component non-linear model according to some embodiments of the present disclosure. The method 900 may be performed by an encoder (e.g., process 200A of FIG. 2A or 200B of FIG. 2B) or a decoder (e.g., process 300A of FIG. 3A or 300B of FIG. 3B) when predicting chroma samples, or may be performed by a plurality of software or hardware components of one device (e.g., device 400 of FIG. 4). For example, one or more processors (e.g., processor 402 of FIG. 4) may perform the method 900. In some embodiments, the method 900 may be implemented by a computer program product, which is embodied in a computer-readable medium including computer-executable instructions such as program code to be executed by a computer (e.g., device 400 of FIG. 4). As shown in FIG. 9, the method 900 includes the following steps 910 to 920.

[0199]

[0209] In step 910, a processor (e.g., processor 402 of FIG. 4) trains an inter-component non-linear model (CCNLM). The CCNLM defines a non-linear relationship between a predicted chroma sample and a concatenated reconstructed luma sample. For example, the CCNLM may be the non-linear model described by Equation 40. The model parameters (i.e., a n , a n-1 ,... a1, a0) in Equation 40 may be arbitrary values. For example, at least one of the parameters a n , a n-1 ,... a1, a0 may be equal to zero. In some embodiments, the model parameters a n , a n-1 ,... a1, a0 are determined based on the bit depth of the picture so that all model parameters are of a similar order of magnitude. In the case of a non-4:4:4 color format (e.g., when the color format is 4:2:2 or 4:2:0),

Number

[0200]

[0210] During model training, the processor uses the training dataset to derive the model parameters a n , a n-1 ,... a1, a0. In some embodiments, the training can be performed by an encoder, and the training dataset includes the original chroma samples of the encoding block and the original concatenated luma samples of the encoding block. After the training is completed, the encoder can encode the derived model parameters a n , a n-1 ,... a1, a0 in the bitstream transmitted to the decoder. In some embodiments, the training dataset includes the reconstructed adjacent chroma samples of the encoding block and the reconstructed adjacent luma samples of the encoding block. Such training based on the reconstructed adjacent luma samples or chroma samples can be performed either on the encoder side or the decoder side, and thus the derived model parameters a n , a n-1,...does not require explicit signaling of a1, a0. In some embodiments, training can be performed by minimizing the mean squared error (MSE) between the reconstructed adjacent luma samples and the predicted adjacent chroma samples of the chroma block. In some embodiments, training can be performed by using the CCLM_LT mode, the CCLM_L mode, or the CCLM_T mode. The CCLM_LT mode uses both the reconstructed luma / chroma samples adjacent above and the reconstructed luma / chroma samples adjacent to the left as training data (e.g., FIGS. 5A, 6A). The CCLM_L mode uses the reconstructed luma / chroma samples adjacent to the left as training data (e.g., FIGS. 5B, 6B). The CCLM_T mode uses the reconstructed luma / chroma samples adjacent above as training data (e.g., FIGS. 5C, 6C).

[0201]

[0211] Referring back to FIG. 9, in step 920, the processor predicts chroma samples based on the CCNLM. In some embodiments, two or more non - linear models may be used to predict chroma samples within an encoding block. For example, the processor may compare the reconstructed luma samples within the encoding block with one or more thresholds, and based on this comparison, classify the reconstructed luma samples within the encoding block into multiple classes. The thresholds may be statistical values of the reconstructed luma samples within the encoding block (such as the average or median of the reconstructed luma samples). Next, the processor may apply a plurality of different non - linear models to each of the multiple classes. In some embodiments, such multi - model training may be performed by using the MMNLM_LT mode, the MMNLM_L mode, or the MMNLM_T mode. The MMNLM_LT mode uses both the reconstructed luma / chroma samples adjacent above and the reconstructed luma / chroma samples adjacent to the left as training data (e.g., FIGS. 5A, 6A). The MMNLM_L mode uses the reconstructed luma / chroma samples adjacent to the left as training data (e.g., FIGS. 5B, 6B). The MMNLM_T mode uses the reconstructed luma / chroma samples adjacent above as training data (e.g., FIGS. 5C, 6C). In some embodiments, the non - linear model used in method 900 may also include terms represented by the gradient of the concatenated reconstructed luma samples. For example, the gradient - based non - linear model may be defined as one of equations 46 - 53. Details for combining the non - linear model and the gradient are described above in relation to these equations.

[0202]

[0212] Figure 10 is a flowchart of a method 1000 for predicting a chroma sample by using a cross-component non-linear model according to some embodiments of the present disclosure. The method 1000 can be performed by an encoder (e.g., process 200A of FIG. 2A or 200B of FIG. 2B) or a decoder (e.g., process 300A of FIG. 3A or 300B of FIG. 3B) when predicting a chroma sample, or can be performed by a plurality of software or hardware components of one device (e.g., device 400 of FIG. 4). For example, one or more processors (e.g., processor 402 of FIG. 4) can perform the method 1000. In some embodiments, the method 1000 can be implemented by a computer program product, which is embodied in a computer-readable medium including computer-executable instructions such as program code executed by a computer (e.g., device 400 of FIG. 4). As shown in FIG. 10, the method 1000 includes the following steps 1010 to 1020.

[0203]

[0213] In step 1010, a processor (e.g., processor 402 of FIG. 4) trains a gradient model. In some embodiments, the gradient model can be a linear model (e.g., Equation 22, Equation 54) that defines a linear relationship between a predicted chroma sample and a concatenated reconstructed luma sample. In some embodiments, the gradient model can be a non-linear model (e.g., Equation 46) that defines a non-linear relationship between a predicted chroma sample and a concatenated reconstructed luma sample. In some embodiments, the gradient model can be a hybrid model (e.g., Equation 39, Equation 49) based on both the gradient of the concatenated reconstructed luma sample and the value of the concatenated reconstructed luma sample. In some embodiments, the gradient model can be a model (e.g., Equation 62) that defines a relationship between a predicted chroma sample of a concatenated reconstructed luma sample and two or more gradients, and each of the one or more gradients is determined by applying different gradient patterns to a plurality of reconstructed luma samples related to the concatenated reconstructed luma sample. For example, the gradient pattern can be a pattern defined by Equations 23 to 38).

[0204]

[0214] During model training, the processor uses a training dataset to derive model parameters. In some embodiments, the training dataset includes one or more rows of samples adjacent to one or more of the current encoding blocks (e.g., FIG. 7A). In some embodiments, the training dataset includes one or more rows of samples adjacent to the upper right of the encoding block (e.g., FIGS. 7C, 8A, 8C). In some embodiments, the training dataset includes one or more rows of samples adjacent to the left of the encoding block (e.g., FIG. 7A). In some embodiments, the training dataset includes one or more rows of samples adjacent to the lower left of the encoding block (e.g., FIGS. 7B, 8A, 8B). In some embodiments, the training dataset includes one or more samples adjacent to the upper left of one or more of the encoding blocks.

[0205]

[0215] Referring back to FIG. 10, in step 1020, the processor predicts chroma samples based on a gradient model. In some embodiments, two or more gradient models may be used to predict chroma samples within an encoding block. For example, the processor may compare the reconstructed luma samples within the encoding block with one or more thresholds, and based on this comparison, classify the reconstructed luma samples within the encoding block into a first plurality of classes. The threshold may include the gradient of the reconstructed luma samples, the sign of the gradient of the reconstructed luma samples, or the absolute value of the downsampling value of the reconstructed luma samples. Next, the processor may apply a plurality of different gradient models to each of the first plurality of classes. In some embodiments, such multi-model training may be performed by using thresholds for classifying the reconstructed adjacent luma and chroma samples into a second plurality of classes. Next, the plurality of gradient models may be trained by using each of the second plurality of classes.

[0206]

[0216] In some embodiments, a non-transitory computer-readable storage medium is also provided. In some embodiments, the medium may store all or a portion of a video bitstream encoded or decoded according to the disclosed cross-component prediction method. Further, the video bitstream may include a flag or a syntax element that signals the disclosed cross-component model or gradient model. For example, the video bitstream may include a flag indicating whether the CCNLM or the gradient model is enabled. As another example, the video bitstream may include a syntax element that signals whether two or more non-linear models or gradient models are used to predict chroma samples in an encoded block from concatenated reconstructed luma samples.

[0207]

[0217] In some embodiments, the non-transitory computer-readable storage medium can store instructions executable by a device (such as the disclosed encoder and decoder) to perform the above method. General forms of non-transitory media include, for example, floppy (registered trademark) disks, flexible disks, hard disks, solid state drives, magnetic tapes, or any other magnetic data storage medium, CD-ROMs, any other optical data storage medium, any physical medium having a pattern of holes, RAM, PROM, and EPROM, FLASH (registered trademark)-EPROM, or any other flash memory, NVRAM, caches, registers, any other memory chip or cartridge, and networked versions of these. The device may include one or more processors (CPUs), an input / output interface, a network interface, and / or memory.

[0208]

[0218] In some embodiments, a computer program product including computer program instructions is provided, and the computer program instructions enable a computer to perform the steps of the method described in any of some embodiments of the present disclosure.

[0209]

[0219] In some embodiments, a computer program is provided that enables a computer to execute the steps of the methods described in any of some embodiments of the present disclosure.

[0210]

[0220] The embodiments may be further described by using the following clauses.

[0211]

[0221] 1. Determining a first value associated with a chroma sample by applying a first gradient pattern to a reconstructed value of a first plurality of luma samples; Determining a second value associated with the chroma sample by applying a downsampling filter to a reconstructed value of a second plurality of luma samples; Predicting the chroma sample based on the first value and the second value and including a video processing method.

[0212]

[0222] 2. The first plurality of luma samples and the second plurality of luma samples are concatenated luma samples of the chroma sample, or adjacent luma samples of the concatenated luma samples each including at least one of the method according to clause 0.

[0213]

[0223] 3. Applying the first gradient pattern to the reconstructed value of the first plurality of luma samples includes encoding or decoding a syntax element in a bitstream, selecting the first gradient pattern from a plurality of gradient patterns based on the syntax element and including the method according to clause 0.

[0214]

[0224] 4. Predicting the chroma sample based on the first value and the second value includes predicting the chroma sample based on a first parameter associated with the first value, a second parameter associated with the second value, and a third parameter, the method according to clause 0.

[0215]

[0225] 5. The prediction of the chroma sample is

Number

Number

[0216]

[0226] 6. The third parameter is determined based on the bit depth of the picture, of the method described in clause 0.

[0217]

[0227] 7. The chroma sample belongs to an encoding block, and the method further includes determining the first parameter, the second parameter, and the third parameter based on one or more adjacent samples of the encoding block, of the method described in clause 0.

[0218]

[0228] 8. One or more adjacent samples of the encoding block are adjacent luma samples of the encoding block, or adjacent chroma samples of the encoding block including at least one of, of the method described in clause 0.

[0219]

[0229] 9. One or more adjacent samples of the encoding block are selected from one or more rows of adjacent samples above or to the left of the encoding block, of the method described in clause 0.

[0220]

[0230] 10. Encoding or decoding a syntax element in a bitstream, determining the number of one or more lines based on the syntax element, and further comprising: when the syntax element is equal to a third value, the number of one or more lines is 1, and when the syntax element is equal to a fourth value, the number of one or more lines is greater than 1, the method according to clause 0.

[0221]

[0231] 11. Encoding or decoding a syntax element in a bitstream, determining one or more adjacent samples of an encoding block based on the syntax element further comprising: when the syntax element is equal to a third value, the one or more adjacent samples are selected from the samples adjacent above and adjacent to the left of the encoding block, when the syntax element is equal to a fourth value, the one or more adjacent samples are selected from the samples adjacent above the encoding block, and when the syntax element is equal to a fifth value, the one or more adjacent samples are selected from the samples adjacent to the left of the encoding block, the method according to clause 0.

[0222]

[0232] 12. Predicting chroma samples based on a first value and a second value includes predicting the Cb and Cr components of the chroma samples based on the first value and the second value, the method according to clause 0.

[0223]

[0233] 13. Predicting chroma samples further includes predicting chroma samples based on a plurality of values, each of the plurality of values being determined by applying different gradient patterns to the reconstructed values of a first plurality of luma samples, the method according to clause 0.

[0224]

[0234] 14. further comprising encoding or decoding a syntax element in a bitstream, The syntax element is the method according to clause 0 indicating whether the prediction of the chroma sample is based on both the first and second values.

[0225]

[0235] 15. A video processing apparatus including a memory for storing a set of instructions, one or more processors wherein the one or more processors cause the apparatus to determine a first value associated with a chroma sample by applying a first gradient pattern to the reconstruction values of a first plurality of luma samples; determine a second value associated with a chroma sample by applying a downsampling filter to the reconstruction values of a second plurality of luma samples; predict the chroma sample based on the first value and the second value; and is configured to execute a set of instructions to perform the above operations.

[0226]

[0236] 16. The first plurality of luma samples and the second plurality of luma samples are concatenated luma samples of the chroma sample, or adjacent luma samples of the concatenated luma samples The apparatus according to clause 0, each including at least one of the above.

[0227]

[0237] 17. In determining a first gradient pattern for the reconstruction values of the first plurality of luma samples, the one or more processors cause the apparatus to encode or decode a syntax element in a bitstream; select a first gradient pattern from a plurality of gradient patterns based on the syntax element; The apparatus according to clause 0, which is configured to execute a set of instructions to perform the above operations.

[0228]

[0238] In predicting chroma samples based on a first value and a second value, one or more processors cause the apparatus to predict chroma samples based on a first parameter associated with the first value, a second parameter associated with the second value, and a third parameter by executing a set of instructions, the apparatus according to clause 0, configured to cause the above.

[0229]

[0239] 19. The prediction of chroma samples is

Number

Number

[0230]

[0240] 20. The third parameter is determined based on the bit depth of the picture, the apparatus according to clause 0.

[0231]

[0241] 21. The chroma samples belong to an encoding block, and one or more processors cause the apparatus to determine the first parameter, the second parameter, and the third parameter based on one or more adjacent samples of the encoding block by executing a set of instructions, the apparatus according to clause 0, configured to cause the above.

[0232]

[0242] One or more adjacent samples of the symbolization block are adjacent luma samples of the symbolization block, or adjacent chroma samples of the symbolization block The apparatus according to clause

[0241] , including at least one of them.

[0233]

[0243] 23. One or more adjacent samples of the symbolization block are selected from one or more rows of samples adjacent above or adjacent to the left of the symbolization block. The apparatus according to clause

[0241] .

[0234]

[0244] 24. One or more processors cause the apparatus to encode or decode syntax elements in the bitstream, and determine the number of one or more rows based on the syntax elements configured to execute a set of instructions to cause, when the syntax element is equal to a third value, the number of one or more rows is 1, and when the syntax element is equal to a fourth value, the number of one or more rows is greater than 1. The apparatus according to clause

[0243] .

[0235]

[0245] 25. One or more processors cause the apparatus to encode or decode syntax elements in the bitstream, determine one or more adjacent samples of the symbolization block based on the syntax elements configured to execute a set of instructions to cause, when the syntax element is equal to a third value, one or more adjacent samples are selected from samples adjacent above and adjacent to the left of the symbolization block, when the syntax element is equal to a fourth value, one or more adjacent samples are selected from samples adjacent above the symbolization block, and The apparatus according to clause

[0241] , wherein when the syntax element is equal to a fifth value, one or more adjacent samples are selected from the samples adjacent to the left of the coded block.

[0236]

[0246] 26. In predicting chroma samples based on a first value and a second value, one or more processors cause the apparatus to predict the Cb and Cr components of the chroma samples based on the first value and the second value by executing a set of instructions, the apparatus according to clause 0.

[0237]

[0247] 27. In predicting chroma samples, one or more processors cause the apparatus to predict chroma samples based on a plurality of values by executing a set of instructions, configured such that each of the plurality of values is determined by applying a different gradient pattern to the reconstructed values of a first plurality of luma samples, the apparatus according to clause 0.

[0238]

[0248] 28. One or more processors cause the apparatus to encode or decode a syntax element in a bitstream by executing a set of instructions, configured such that the syntax element indicates whether the prediction of the chroma sample is based on both the first and second values, the apparatus according to clause 0.

[0239]

[0249] 29. A non-transitory computer-readable medium storing a bitstream of video for processing by a method, the method comprising: determining a first value associated with a chroma sample by applying a first gradient pattern to the reconstructed values of a first plurality of luma samples; and determining a second value associated with a chroma sample by applying a downsampling filter to the reconstructed values of a second plurality of luma samples. Predicting chroma samples based on a first value and a second value A non - transitory computer - readable medium including the above.

[0240]

[0250] 30. The first plurality of luma samples and the second plurality of luma samples are either concatenated luma samples of chroma samples, or adjacent luma samples of the concatenated luma samples, The non - transitory computer - readable medium according to clause

[0249] , each including at least one of the above.

[0241]

[0251] 31. Applying a first gradient pattern to a reconstructed value of the first plurality of luma samples is encoding or decoding a syntax element in a bitstream, and selecting the first gradient pattern from a plurality of gradient patterns based on the syntax element The non - transitory computer - readable medium according to clause

[0249] , including the above.

[0242]

[0252] 32. Predicting chroma samples based on a first value and a second value is predicting chroma samples based on a first parameter associated with the first value, a second parameter associated with the second value, and a third parameter The non - transitory computer - readable medium according to clause

[0249] , including the above.

[0243]

[0253] 33. The prediction of chroma samples is

Number

Number

[0252] .

[0244]

[0254] 34. The third parameter is determined based on the bit depth of the picture, for the non-transitory computer-readable medium described in clause

[0252] .

[0245]

[0255] 35. The chroma sample belongs to an encoding block, and the method further includes determining the first parameter, the second parameter, and the third parameter based on one or more adjacent samples of the encoding block, for the non-transitory computer-readable medium described in clause

[0252] .

[0246]

[0256] 36. One or more adjacent samples of the encoding block are adjacent luma samples of the encoding block, or adjacent chroma samples of the encoding block including at least one of them, for the non-transitory computer-readable medium described in clause

[0255] .

[0247]

[0257] 37. One or more adjacent samples of the encoding block are selected from one or more rows of samples adjacent above or adjacent to the left of the encoding block, for the non-transitory computer-readable medium described in clause

[0255] .

[0248]

[0258] 38. Encoding or decoding a syntax element in a bitstream, and determining the number of one or more rows based on the syntax element further includes, when the syntax element is equal to the third value, the number of one or more rows is 1, and The non - transitory computer - readable medium according to clause

[0257] , where when the syntax element is equal to a fourth value, the number of one or more lines is greater than 1.

[0249]

[0259] 39. Encoding or decoding a syntax element in a bitstream, determining one or more adjacent samples of an encoded block based on the syntax element further comprising, when the syntax element is equal to a third value, one or more adjacent samples are selected from the samples adjacent above and adjacent to the left of the encoded block, when the syntax element is equal to a fourth value, one or more adjacent samples are selected from the samples adjacent above the encoded block, and when the syntax element is equal to a fifth value, one or more adjacent samples are selected from the samples adjacent to the left of the encoded block, the non - transitory computer - readable medium according to clause

[0255] .

[0250]

[0260] 40. Predicting chroma samples based on a first value and a second value includes predicting the Cb component and the Cr component of the chroma samples based on the first value and the second value, the non - transitory computer - readable medium according to clause

[0249] .

[0251]

[0261] 41. Predicting chroma samples further includes predicting chroma samples based on a plurality of values, each of the plurality of values is determined by applying different gradient patterns to the reconstructed values of a first plurality of luma samples, the non - transitory computer - readable medium according to clause

[0249] .

[0252]

[0262] 42. further comprising encoding or decoding a syntax element in a bitstream, the syntax element indicating whether the prediction of the chroma sample is based on both the first and second values, the non - transitory computer - readable medium according to clause

[0249] .

[0253]

[0263] 43. A video processing method including predicting a chroma sample from a concatenated luma sample related to the chroma sample, the prediction being based on a non-linear model defining a non-linear relationship between a predicted value of the chroma sample and a value related to the concatenated luma sample.

[0254]

[0264] 44. The non-linear model is of the non-linear form:

Number

Number

[0263] .

[0255]

[0265] 45. At least one of the parameters a n ,a n-1 ,...a1,a0 is equal to zero, the method according to clause

[0264] .

[0256]

[0266] 46. At least one of the parameters a n ,a n-1 ,...a1,a0 is determined based on the bit depth of the picture, the method according to clause

[0264] .

[0257]

[0267] 47. The chroma sample belongs to an encoding block, and the method is based on adjacent samples of the encoding block to determine the parameters a n ,an-1 ,... further including determining a1, a0, the method according to clause

[0264] .

[0258]

[0268] 48. The method according to clause

[0263] , further including determining a value associated with a concatenated chroma sample by applying a downsampling filter to a reconstructed value of a plurality of luma samples related to the concatenated luma sample.

[0259]

[0269] 49. Two or more non - linear models are used to predict chroma samples within an encoding block, and the method includes classifying luma samples within the encoding block into a plurality of classes, and applying a plurality of non - linear models to the plurality of classes respectively and further including the method according to clause

[0263] .

[0260]

[0270] 50. The non - linear model further defines a relationship between a predicted value of a chroma sample and a gradient value related to a concatenated luma sample, the gradient value is determined based on applying a gradient pattern to a reconstructed value of a plurality of luma samples related to the concatenated luma sample, the method according to clause

[0263] .

[0261]

[0271] 51. The non - linear model includes at least one term represented by a n ·(G l (i, j)) n where, in the formula, (i, j) represents the coordinates of the chroma sample, G L (i, j) is the gradient value, n is an integer of 1 or more, a n is a parameter of the non - linear model, the method according to clause

[0270] .

[0262]

[0272] 52. The non - linear model

Number

Number

[0271] .

[0263]

[0273] 53. A video processing apparatus including a memory for storing a set of instructions and one or more processors, one or more processors, wherein the one or more processors are configured to execute the set of instructions to cause the apparatus to predict chroma samples from the concatenated luma samples related to the chroma samples, and the prediction is based on a non-linear model that defines a non-linear relationship between the predicted value of the chroma samples and the value related to the concatenated luma samples.

[0264]

[0274] 54. The non-linear model is in the non-linear form:

Number

Number

[0273] .

[0265]

[0275] 55. Parameter a n , a n-1 ,... at least one of a1, a0 is equal to zero, the apparatus according to clause

[0274] .

[0266]

[0276] 56. Parameter a n , a n-1 ,... at least one of a1, a0 is determined based on the bit depth of the picture, the apparatus according to clause

[0274] .

[0267]

[0277] 57. The chroma sample belongs to an encoding block, and one or more processors cause the apparatus to execute a set of instructions to determine parameter a n , a n-1 ,... a1, a0 based on adjacent samples of the encoding block, the apparatus according to clause

[0274] .

[0268]

[0278] 58. One or more processors cause the apparatus to execute a set of instructions to determine a value associated with a concatenated chroma sample by applying a downsampling filter to a reconstructed value of a plurality of luma samples associated with the concatenated luma samples, the apparatus according to clause

[0273] .

[0269]

[0279] 59. Two or more non - linear models are used to predict chroma samples within an encoding block, one or more processors cause the apparatus to classify luma samples within the encoding block into a plurality of classes and apply a plurality of non - linear models to the plurality of classes respectively and cause the apparatus to execute a set of instructions for doing so, the apparatus according to clause

[0273] .

[0270]

[0280] 60. The non-linear model further defines the relationship between the predicted value of the chroma sample and the gradient value associated with the concatenated luma samples, The gradient value is determined based on applying a gradient pattern to the reconstructed values of a plurality of luma samples associated with the concatenated luma samples, the apparatus according to clause

[0273] .

[0271]

[0281] 61. The non-linear model is a n ·(G l (i,j)) n including at least one term represented by, where in the formula, (i,j) represents the coordinates of the chroma sample, G L (i,j) is the gradient value, n is an integer greater than or equal to 1, a n is a parameter of the non-linear model, the apparatus according to clause

[0280] .

[0272]

[0282] 62. The non-linear model is

Number

Number

[0281] .

[0273]

[0283] 63. A non-transitory computer-readable medium storing a bitstream of a video for processing by a method including predicting a chroma sample from concatenated luma samples related to the chroma sample, A non - transitory computer - readable medium, where the prediction is based on a non - linear model that defines a non - linear relationship between the predicted value of a chroma sample and a value associated with a concatenated luma sample.

[0274]

[0284] 64. The non - linear model is in the non - linear form:

Number

Number

[0283] .

[0275]

[0285] 65. At least one of the parameters a n ,a n-1 ,...a1,a0 is equal to zero, the non - transitory computer - readable medium described in clause

[0284] .

[0276]

[0286] 66. At least one of the parameters a n ,a n-1 ,...a1,a0 is determined based on the bit - depth of the picture, the non - transitory computer - readable medium described in clause

[0284] .

[0277]

[0287] 67. The chroma sample belongs to an encoding block, and the method is based on adjacent samples of the encoding block for the parameters a n ,a n-1,...a1, a0, further including determining, the non-transitory computer-readable medium according to clause

[0284] .

[0278]

[0288] 68. The non-transitory computer-readable medium according to clause

[0283] , further including determining a value associated with a concatenated chroma sample by applying a downsampling filter to a reconstruction value of a plurality of luma samples related to the concatenated luma sample.

[0279]

[0289] 69. Two or more non-linear models are used to predict chroma samples in an encoding block, and the method includes classifying luma samples in the encoding block into a plurality of classes, and applying a plurality of non-linear models to the plurality of classes respectively and further including, the non-transitory computer-readable medium according to clause

[0283] .

[0280]

[0290] 70. The non-linear model further defines a relationship between a predicted value of a chroma sample and a gradient value related to a concatenated luma sample, and the gradient value is determined based on applying a gradient pattern to a reconstruction value of a plurality of luma samples related to the concatenated luma sample, the non-transitory computer-readable medium according to clause

[0283] .

[0281]

[0291] 71. The non-linear model includes at least one term represented by a n ·(G l (i,j)) n where in the formula, (i,j) represents the coordinates of the chroma sample, G L (i,j) is the gradient value, n is an integer of 1 or more, a n is a parameter of the non-linear model, the non-transitory computer-readable medium according to clause

[0290] .

[0282]

[0292] 72. The non-linear model further includes at least one term represented by [Math] where, in the formula, [Math] is a value related to the concatenated lum samples, m is an integer greater than or equal to 1, a m is a parameter of the non-linear model, the non-transitory computer-readable medium according to clause

[0291] .

[0283]

[0293] 73. A computer program product including computer program instructions, the computer program instructions enabling a computer to execute the video processing method according to any one of clauses 1 to 14 and clauses 43 to 52.

[0284]

[0294] 74. A computer program enabling a computer to execute the video processing method according to any one of clauses 1 to 14 and clauses 43 to 52.

[0285]

[0295] Note that relative terms in this specification such as "first" and "second" are merely used to distinguish an entity or operation from another entity or operation, and do not require or imply any actual relationship or order between these entities or operations. Further, the words "comprising", "having", "containing", "including" and other similar forms have equivalent meanings, and the elements or groups of elements following any of these words are not meant to be a limiting enumeration of such elements or groups of elements, or are not meant to be limited only to the enumerated elements or groups of elements, and are intended to be open-ended.

[0286]

[0296] As used herein, unless otherwise specifically stated, the term "or" includes all possible combinations, except when it is not feasible. For example, if it is stated that a database may include A or B, then, unless otherwise specifically stated or it is not feasible, the database may include A, B, or both A and B. As a second example, if it is stated that a database may include A, B, or C, then, unless otherwise specifically stated or it is not feasible, the database may include A, B, C, A and B, A and C, B and C, or A, B, and C.

[0287]

[0297] It is understood that the above-described embodiments may be implemented by hardware, software (program code), or a combination of hardware and software. When implemented by software, it may be stored in the above-described computer-readable medium. The software can execute the method of the present disclosure when executed by a processor. The computing units and other functional units described in the present disclosure may be implemented by hardware, software, or a combination of hardware and software. Those skilled in the art will also understand that a plurality of the above-described modules / units may be combined into one module / unit, and each of the above-described modules / units may be further divided into a plurality of sub-modules / sub-units.

[0288]

[0298] In the above description, the embodiments have been described with reference to many specific details that may vary for each implementation form. Specific adaptation forms and modification forms of the above-described embodiments may be made. From the consideration of this specification and the implementation of the invention disclosed in this specification, other embodiments may become apparent to those skilled in the art. This specification and the examples are intended to be considered only as examples, and the true scope and spirit of the invention are indicated by the appended claims. It is intended that the sequence of steps shown in the figures is for illustrative purposes only and is not intended to be limited to any specific sequence of steps. Therefore, those skilled in the art can understand that these steps may be executed in a different order while implementing the same method.

[0289]

[0299] Drawings and exemplary embodiments have been disclosed in the specification. However, many variations and modifications can be made to these embodiments. Therefore, even if specific terms are adopted, they are merely used in the sense of general description and not for the purpose of limitation.

Claims

1. Decoding the bitstream containing the first flag, In response to the first flag having a first value, determine the gradient of the reconstructed luma block associated with the encoded block, and based on the gradient, predict the concatenated chroma block of the reconstructed luma block, or In response to the first flag having a second value, downsample the reconstructed luma block and predict the connected chroma block based on a linear combination of the gradient and the downsampled reconstructed luma block, wherein the gradient and the downsampled reconstructed luma block in the linear combination each have different linear parameters. A video decoding method, including the above.

2. The method according to claim 1, wherein the encoding block has a non-4:4:4 color format.

3. The bitstream includes a second flag, and the method is The method according to claim 1, further comprising determining, based on the value of the second flag, whether the gradient of the reconstructed Luma block is used to predict both the Cr and Cn components of the coding block.

4. The bitstream includes a second flag, and the method is The method according to claim 1, further comprising determining, based on the value of the second flag, whether the gradient of the reconstructed luma block is used to predict the connected chroma block.

5. Decoding the syntax elements of the bitstream, wherein the syntax elements are encoded by a truncated unary code and exhibit a gradient pattern. The gradient of the reconstructed Luma block is determined using the indicated gradient pattern, The method according to claim 1, further comprising:

6. The method according to claim 1, wherein the linear parameter in the linear combination is derived based on chroma samples in a plurality of adjacent rows of the connected chroma blocks.

7. The method according to claim 6, wherein the plurality of adjacent rows include six adjacent rows.

8. When the coded block is coded in CCLM_LT mode, the multiple adjacent rows include the six samples above and the six rows to the left. When the coded block is coded in CCLM_L mode, the plurality of adjacent rows include the six rows on the left and the six rows below the left, or When the coded block is coded in CCLM_T mode, the plurality of adjacent rows include the top six rows and the top right six rows. The method according to claim 6.

9. The aforementioned linear combination is, [Math 1] And in the formula, (i, j) represents the coordinates of the chromatic sample of the linked chromatic block, pred Ｃ (i, j) are the predicted values ​​of the chroma sample, G Ｌ (i, j) is the value of the gradient, [Math 2] is the downsampled and reconstructed luma block value, and a １ a ２ and a ３ The linear parameter of the linear combination is The method according to claim 1.

10. Determining the gradient of the reconstructed Luma block associated with the coded block, To predict the concatenated chromatic block of the reconstructed luma block, Encoding a bitstream associated with the encoding block, wherein the bitstream includes a first flag. If the first flag has a first value, the connected chromablock is predicted based on the gradient. If the first flag has a second value, the reconstructed luma block is downsampled, and the concatenated chroma block is predicted based on a linear combination of the gradient and the downsampled reconstructed luma block, wherein the gradient and the downsampled reconstructed luma block in the linear combination each have different linear parameters. Video encoding method.

11. The method according to claim 10, wherein the encoding block has a non-4:4:4 color format.

12. The method according to claim 10, further comprising encoding a second flag in the bitstream indicating whether the gradient of the reconstructed Luma block is used to predict both the Cr and Cn components of the coded block.

13. The method according to claim 10, further comprising encoding a second flag in the bitstream indicating whether the gradient of the reconstructed luma block is used to predict the concatenated chroma block.

14. Selecting a gradient pattern from multiple gradient patterns to be used to determine the gradient of the reconstructed Luma block, The bitstream is encoded with a syntax element representing the selected gradient pattern, wherein the syntax element is encoded using a shortened unary code. The method according to claim 10, further comprising:

15. The method according to claim 10, wherein the linear parameter in the linear combination is derived based on chroma samples in a plurality of adjacent rows of the connected chroma blocks.

16. The method according to claim 15, wherein the plurality of adjacent rows include six adjacent rows.

17. When the coded block is coded in CCLM_LT mode, the multiple adjacent rows include the six samples above and the six rows to the left. When the coded block is coded in CCLM_L mode, the plurality of adjacent rows include the six rows on the left and the six rows below the left, or When the coded block is coded in CCLM_T mode, the plurality of adjacent rows include the top six rows and the top right six rows. The method according to claim 15.

18. The aforementioned linear combination is, [Math 1] And in the formula, (i, j) represents the coordinates of the chromatic sample of the linked chromatic block, pred Ｃ (i, j) are the predicted values ​​of the chroma sample, G Ｌ (i, j) is the value of the gradient, [Math 2] is the downsampled and reconstructed luma block value, and a １ 、 a ２ and a ３ are the linear parameters of the linear combination, The method according to claim 10.

19. A method for storing bitstreams associated with one or more pictures, Determining the gradient of the reconstructed Luma block associated with the coded block, A bitstream is generated that includes a first flag indicating a method for predicting the concatenated chroma block of the reconstructed luma block, This includes storing the bitstream in a non-temporary computer-readable medium, If the first flag has a first value, the connected chromablock is predicted based on the gradient. If the first flag has a second value, the reconstructed luma block is downsampled, and the concatenated chroma block is predicted based on a linear combination of the gradient and the downsampled reconstructed luma block, wherein the gradient and the downsampled reconstructed luma block in the linear combination each have different linear parameters. method.

20. The method according to claim 19, wherein the encoding block has a non-4:4:4 color format.

Images