Cross-component prediction tool for video coding

Abstract

This paper describes methods and systems for encoding and decoding video using cross-component prediction tools. The proposed methods include interlumer-based modeling for chroma, cross-component residual prediction, and convolutional cross-component intra prediction.

Description

[Technical Field]

[0001] This document generally relates to the coding of images and videos. More specifically, embodiments of the present invention relate to the application of a cross-component prediction tool in video coding. [Background technology]

[0002] In 2020, the MPEG group within the International Standardization Organization (ISO), in collaboration with the International Telecommunications Union (ITU), released the first edition of the Versatile Video Coding Standard (VC), known as H.266 (reference [1]). More recently, the group has been working on developing a next-generation coding standard that offers improved coding performance compared to existing video coding technologies. As part of this research, new coding technologies are also being tested.

[0003] As acknowledged by the inventors herein, improvements in the techniques for applying cross-component prediction tools in image and video coding are desired and described herein. As used herein, the term “cross-component prediction” refers to predicting lumen pixel values ​​and / or chroma pixel values ​​using chroma and lumen-chroma, respectively. That is, lumen prediction includes chroma values, and chroma prediction includes lumen values.

[0004] The approaches described in this section are approaches that may be pursued, but not necessarily approaches that have been conceived or pursued previously. Therefore, unless otherwise indicated, none of the approaches described in this section should be considered prior art simply because they are included in this section. Similarly, any problems identified with respect to one or more approaches should not be assumed to be recognized in any prior art based on this section, unless otherwise indicated.

[0005] Embodiments of the present invention are shown in the accompanying drawings as examples, not as limitations. In the drawings, the same reference numerals refer to similar elements. [Brief explanation of the drawing]

[0006] [Figure 1] This shows an example of the pixel sample positions used in a cross-component linear model (CCLM). [Figure 2] This shows an example of spatial weights for a convolutional filter used in CCLM. [Figure 3] This shows an example of a reference area (with padding) used in CCLM. [Figure 4] This shows an example of a Sobel-based filter used in a gradient linear model (GLM). [Figure 5] This shows an example of ruma-based modeling of chroma in an intracoded coding unit (CU), following prior art. [Figure 6] An example of chroma-based modeling of an interCU chroma is shown according to an embodiment of the present invention. [Figure 7] This describes an exemplary process of the data flow of an inter-LMC according to an embodiment of the present invention. [Figure 8A] This illustrates an exemplary process of dataflow for cross-component residual modeling in an encoder, according to an embodiment of the present invention. [Figure 8B]This illustrates an exemplary process of cross-component residual modeling in a decoder, according to embodiments of the present invention. [Figure 9] An example of a reference area used to implicitly derive model parameters for cross-component residual modeling (CCRM) according to an embodiment of the present invention is shown. [Modes for carrying out the invention]

[0007] This specification describes exemplary embodiments relating to the application of a cross-component prediction tool in video coding. In the following description, numerous specific details are provided for illustrative purposes to ensure a full understanding of the various embodiments of the invention. However, it will be apparent that various embodiments of the invention may be carried out without these specific details. In other instances, well-known structures and devices are not described in comprehensive detail to avoid unnecessarily obscuring, obscuring, or making embodiments of the invention difficult to understand.

[0008] [summary] The exemplary embodiments described herein relate to the application of cross-component prediction tools in image and video coding for intra-prediction or inter-prediction. The proposed methods include 1) interlumer-based modeling for chroma (interlumer and intralumer-chroma for intercoding units (CUs)), 2) cross-component residual prediction, and 3) convolutional cross-component intra-prediction.

[0009] [Cross-component prediction in video coding] In enhanced compression model (ECM) software implementations, such as ECM7 and later (reference [2]), the cross-component linear model (CCLM) prediction mode, the convolutional cross-component intra prediction model (CCCM) mode, and the gradient linear model (GLM) mode predict chroma using lumen reconstruction pixel modeling. A brief overview of these prediction modes follows.

[0010] [CCLM: Cross-component linear model prediction mode] In VVC (reference [1]), the CCLM mode predicts chroma samples based on reconfigured luma samples of the same coding unit (CU) by using a linear model as follows: Nod C (i,j)=α·rec' L (i,j)+β (1) Here, pred C (i,j) represents the predicted chroma sample within CU, and rec' L (i,j) represents a downsampled reconstructed lumen sample of the same CU, and α and β represent the model parameters.

[0011] It has three modes: linear model (LM) mode (which uses both the upper neighbor sample and the left neighbor sample), LM-A mode (which uses only the upper neighbor sample), and LM-L mode (which uses only the left neighbor sample).

[0012] The parameters of α and β are estimated as follows. FIG. 1 shows an example of the positions of the left sample and the upper sample involved in the CCLM mode compared with the sample of the current block. The group of four adjacent luma samples at the selected positions is downsampled to find two larger values x 0 A and x 1 A and two smaller values x 0 B and x 1 B by comparing four times. Their corresponding chroma sample values (Rec C ) are represented as y 0 A , y 1 A , y 0 B , and y 1 B . Then, X a , X b , Y a , and Y b are derived as follows:[[ID=3b]] X a =(x 0 A +x 1 A +1)>>1 X b =(x 0 B +x 1 B +1)>>1 (2) Y a =(y 0 A +y 1 A +1)>>1 Y b =(y 0 B +y 1 B +1)>>1 Here, x>>n means shifting x n bits to the right in binary, or dividing x by 2 n . Finally, the linear model parameters α and β are obtained according to the following formula: α=(Y a -Y b ) / (X a -X b ) (3) β=Y b -α·X b

[0013] [CCCM: Convolutional cross-component intra prediction model] CCCM is similar to CCLM, but uses a convolutional filter, where the 7-tap filter consists of a 5-tap plus-shaped spatial component, a nonlinear term (P), and a bias term (B), as follows: predChoromaVal =c0C+c1N+c2S+c3E+c4W+c5P+c6B (4) Here, c i Here, for i=0 to 6, the filter coefficients are represented, B is the bias term, and P is a nonlinear term calculated as follows: P=(C*C+midVal)>>bitDepth Here, bitDepth represents the bit depth of the luma component, and midVal=2 (bitDepth-1) That is the case.

[0014] The input to the filter's 5-tap spatial component consists of the predicted chroma sample, the central (C) chroma sample in its collocation, and its upper / north (N), lower / south (S), left / west (W), and right / east (E) neighbors, as shown in Figure 2. The bias term B represents the scalar offset between the input and output and is typically set to the intermediate chroma value (e.g., 512 for 10-bit content). Filter coefficient c iThis is calculated by subtracting the mean squared error (MSE) between the predicted chroma sample and the reconstructed chroma sample within the reference area. For example, as shown in Figure 3, a predicted unit (PU) is surrounded by a reference area (305) consisting of n (e.g., n=6) rows or columns of chroma samples above and to the left of the PU. The reference area extends 1 PU width to the right and 1 PU height below the PU boundary. The reference area is adjusted to include only available samples. The extension to the reference area (1 pixel row or 1 pixel column), shown in black, is necessary to support the "side samples" of the cross-shaped spatial filter shown in Figure 2 and is padded when the filter is applied in an unavailable area.

[0015] [GLM: Gradient Linear Model] GLM utilizes lumens sample gradients G to derive linear models: C = α·G + β (5) Four 3x2 gradient filters are effective in the case of GLM, as shown in Figure 4.

[0016] As will be discussed next, in exemplary embodiments of the present invention, several novel cross-component prediction tools are proposed.

[0017] [Interluma-based modeling for chroma (Interluma and intraLM chroma for interCU)] In past conventional codecs, for the purpose of intercoding tools, the luma and chroma share the same or scaled motion vectors (MV) based on the sampling format of the luma and chroma (e.g., 4:2:0, 4:4:4, etc.). In exemplary embodiments of this disclosure, it is proposed to break these conventional rules. For example, the luma uses MV for inter-prediction, while the chroma uses a luma-dependent cross-component predictive tool. The basic idea is that for a given content, for the intercoding unit, the chroma (505) may be better predicted using a reconstructed luma (510, 520) along with cross-component intra-chroma tools such as CCLM, CCCM, and GLM. Several cross-component predictive tools exist available in ECM[2] for intra-coded CUs, and as shown in Figure 5, the chroma (505) is predicted from a luma-reconstructed sample of the intra-coded CU (510) using modeling parameters derived by techniques such as CCLM, CCCM, and GLM.

[0018] There are certain contents (e.g., high-texture regions) in which the interCU will be coded with more residual bits for chroma, and cross-component prediction may help further improve coding gain through improved chroma prediction. The exemplary embodiments presented herein propose improving compression efficiency by applying cross-component prediction to the interCU.

[0019] As shown in Figure 6, the current luma block (605) is predicted using inter-prediction and reconstructed as an arbitrary inter-CU, while the chroma block (610) is predicted from the current luma reconstruction sample (605) using appropriate modeling parameters derived using techniques such as CCLM, CCCM, GLM, or other similar modeling techniques. This coding tool is called Inter-LMC (Luma based modeling for chroma for Inter-CUs).

[0020] In another embodiment, instead of using reconstructed pixels of adjacent lumens and chromas, prediction samples of lumens and chromas currently derived in intermode of the CU can be used to derive either the CCLM model or the CCCM model, and the model can be applied to the lumens reconstruction of the CU to derive the interLMC predictor.

[0021] In another embodiment, instead of always using reconstructed pixels of adjacent lumens and chromas, either the predicted lumens and chromas derived in the intermode of the CU or the reconstructed pixels of adjacent lumens and chromas can be used to derive either the CCLM model or the CCCM model, and the selection can then be signaled by a new syntax flag.

[0022] In another embodiment, either the CCLM or CCCM model is modified, based on the CU size, to use either reconstructed pixels of adjacent lumens and chromas or predicted samples of lumens and chromas derived from the intermode of the current CU. Model parameters can be derived using either low-level decomposition (LLD) or Gaussian decomposition.

[0023] In another embodiment, either the CCLM or CCCM model is modified to use both reconstructed pixels of adjacent lumens and chromas, and predicted samples of lumens and chromas currently derived from the intermode of the CU. Model parameters can be derived using either low-level decomposition (LLD) or Gaussian decomposition.

[0024] In another embodiment, chroma prediction may be derived using a fusion of a CCLM model and a CCCM model. The fusion weight (w) can be fixed (e.g., w=0.5), or w can be derived by a template matching process. For example, the fusion is: T Ref =w*T Inter +(1-w)*T LMC (6) When expressed as, the minimization model for deriving w is:

number

[0025] In another embodiment, the encoder may select from various chroma prediction modes (see above) and signal the selected mode with appropriate syntax parameters. Examples of such syntaxes are shown in Table 1. [Table 1] An inter_lmc_flag equal to 1 indicates that interLMC will be used. An inter_lmc_flag equal to 0 indicates that interLMC will not be used. inter_lmc_mode_idx specifies the index of the selected inter-LMC mode (for example, a selection between LM_CHROM_IDX and MMLM_CHROMA_IDX). A cccm_flag equal to 1 indicates that CCCM will be used. A cccm_flag equal to 0 indicates that CCLM will be used.

[0026] Figure 7 shows an exemplary data flow for deriving chroma prediction in inter-LMC mode. As shown in Figure 7, the decoder reconstructs pixels based on the intercoding syntax parameters (705). If inter_lmc_flag==0, the decoder uses further syntax elements (such as inter_lmc_mode_idx and cccm_flag) to derive additional parameters for modeling chroma reconstruction (710, 715). The derived model parameters are then applied to derive the reconstructed chroma samples (720).

[0027] Experiments have shown that coding efficiency can be improved by disabling subblock division in the case of inter-LMC, as shown in Table 2. Therefore, in some embodiments, it may be desirable not to use subblock division in the case of inter-LMC. For example, under Class F, the compression gain for BD-Y without block sub-division is -0.18%, compared to -0.01% with subblock division. In Table 2, Class D and Class F refer to the classes of video test sequences used in JVET. [Table 2]

[0028] [Cross-component residual modeling (CCRM) and cross-component residual prediction (CCRP)] This is an intra-prediction tool that aims to achieve better chroma prediction using lumar residual modeling. a. Reconstruction involves both low-frequency and high-frequency signals, and conventional intra-prediction may be helpful in better predicting the low-frequency signals. b. The high-frequency portion can be predicted using a cross-component residual model.

[0029] The method can be described as follows: PredL and ReconL are considered to be the predicted ruma data and reconstructed ruma data, respectively. PredC is the predicted chroma data corresponding to the ruma intra prediction mode, in which case the following holds: PredCfinal=PredC+(alpha*(luma residual)) (7) Here, the alpha value (modeling parameter) is derived by the decoder based on the modeling using the lumar and chromar neighborhood residual characteristics. Alternatively, the modeling parameter may be derived by the encoder and signaled to the decoder via appropriate syntax parameters. In equation (7), the prediction can be improved by using the lumar residual instead of the lumar pixel value. This is because the high-frequency portion of the chromar signal is better predicted when the modeling is improved using only the high-frequency residual signal.

[0030] In this embodiment, the alpha parameter can be calculated using modeling with nearby lumern residuals and chromar residuals. C represents the chroma residual value, and Res' L Assume that represents the downsampled lumer residual value. In that case, alpha(α) can be derived based on linear least squares error estimation using causal residual samples around the current block, as follows:

number

[0031] In another embodiment, alpha can be explicitly signaled from the encoder to the decoder. For example, a set of alpha values ​​such as [0, ±1 / 8, ±1 / 4, ±1 / 2, ±1] may be accepted. In this case, a syntax parameter such as alpha_idx can be used to signal, along with a sign flag, which alpha value within this prespecified table is selected for each chroma component.

[0032] Figure 8A shows an example of the encoder-side data flow for determining modeling CCRM / CCPM parameters, illustrating how chroma residuals are derived using cross-component residual modeling. The process uses the original (incoming) uncompressed chroma value (802), the predicted chroma value (805), and the predicted lumar residual value (803) as they are expected to be received by the decoder (e.g., after inverse quantization and inverse transform, though not shown). Considering the input chroma value (802) and the predicted chroma value (805), the subtractor 810 generates a residual chroma value (808), which, combined with the predicted lumar residual value (830), is used to derive CCRM / CCPM model parameters (e.g., alpha). The output of the "Luma residual modeling" (835), for example, alpha*(luma residual)(837), is further subtracted from the chroma residual (808), and the final chroma residual output (817) (along with the corresponding luma residual, though not shown) is sent to transformation and quantization (820) and entropic coding (840) to generate the output bitstream (827).

[0033] In one embodiment, in the unit “Derivation of Model Parameters” (825), the modeling parameters of CCRM / CCPM may be derived using the lumens and chromens' neighbor residual characteristics. These parameters may be regenerated in the decoder, or alternatively, they may be signaled to the decoder via appropriate syntax parameters by considering the bit-to-distortion trade-off.

[0034] Figure 8B shows the normative data flow for decoding based on CCRM / CCPM. Given the coded bitstream (827), after inverse quantization (IQ) and inverse transform (IT) (850), the lumern residual (854) and chroma residual (852) are extracted. In unit 825B, CCRM modeling parameters (e.g., alpha) are either explicitly derived based on the decoded information, or they are extracted from the bitstream via appropriate syntax parameters. In unit 865, the lumern residual (854) and modeling parameters are applied to derive the lumern modeling residual (867) (e.g., alpha*(lumern residual)), which is added to the chroma residual (852) and the chroma value of the intra-predicted data (862C) to obtain the improved predicted output value of the chroma (877). Similarly, the rumor value (862L) of the intra-predicted data is added to the rumor residual (854) to derive the rumor output value (857).

[0035] In one embodiment, the model parameters / coefficients (e.g., alpha) can be derived by solving the linear equations corresponding to the neighboring lumer residuals and neighboring chromer residuals using LDL decomposition or Gaussian elimination. In another embodiment, neighboring samples such as the left sample, right sample, upper sample, and lower sample can be used in addition to the sample as a five-parameter model, with or without nonlinear or bias terms.

[0036] FIG. 9 shows an example of a neighborhood residual sample area including M up-going and N left columns of chroma (e.g., M = 2, N = 2). The corresponding luma reference residual area is twice the size of the chroma, which is downsampled / subsampled for modeling and used. To enhance the accuracy of the model parameters, continuity checks are performed along the upper and left boundaries using the current luma residual sample and neighboring luma residual samples. For example, only when |(2*R(X, -1) - R(X, -2) - R(X, 0))| < Th, the neighborhood residuals from R(X, -1) to R(X, -M) are considered for modeling. Similarly, only when |2*R(-1, Y) - R(-2, Y) - R(0, Y)| < Th, the neighborhood residuals from R(-1, Y) to R(-N, Y) are considered for modeling. Here, Th represents a threshold for ensuring the continuity of the luma residuals across the CU boundary (e.g., Th = 10 << (Bitdepth - 8)).

[0037] In another embodiment, the model parameters (e.g., alpha) can be explicitly signaled for a particular CU, while for other CUs, these parameters can be inferred from neighboring CUs whose prediction modes and residual characteristics are similar to the current CU. This new mode can be called the CCRM_merge mode. For example, when the intra prediction mode of the current CU is in the "non-LM" mode (i.e., chroma prediction does not use a luma modeling mode such as CCLM, CCCM, or GLM), only neighboring CUs with the non-LM intra prediction mode are considered. Among these neighboring CUs, as an example, the best neighboring CU can be selected by determining the minimum absolute difference between the average of the luma residual samples of the neighboring CU and the average of the current luma residual samples. If the neighboring CU uses CCRM, the alpha model parameter can be directly used from it; otherwise, the model parameter can be derived using its residual. The following pseudo-code shows an example of the use of the proposed CCRM_merge mode.

Number

[0038] In one embodiment, predefined model parameters may be used to reduce the signaling overhead of signaling the sign flag and alpha_index value. Such predefined model parameters may also be signaled once at the header level (e.g., in the picture header or slice header).

[0039] In another embodiment, the CCRM flag may signal a specific chromatin intra-prediction mode. For example, the CCRM prediction mode (and therefore the CCRM flag) may be permitted only for normal intra-prediction modes such as DC, Planar, and Angular, but not for cross-component prediction modes such as CCLM, CCCM, and GLM.

[0040] In another embodiment, model parameters may be implicitly derived using neighboring lumens and chroma prediction data.

[0041] [CCCM improvement] The basic idea is to introduce an additional nonlinear element (e.g., a nonlinear kernel) as input to a 7-tap convolutional filter that replaces the original weighted sum of the central lumern sample (C) and the four neighboring lumern samples (N, S, W, E). The proposed nonlinear CCCM (NL-CCCM) considers not only the strength of the reconstructed lumern samples but also the similarity between the neighboring samples and the central sample. The filter coefficient derivation process remains unchanged (reference [2]).

[0042] In one embodiment, a nonlinear kernel can be used to replace the weighted sum of four neighboring samples. The predicted chroma value can be derived as follows: predChromaVal=c0C+c1K(N)+c2K(S)+C3K(E)+c4K(W)+c5P+c6B (9) Here, K(·) is a nonlinear function that takes into account the similarity and / or strength with the central luma. For X∈in{N,S,E,W}, the kernel function K(X) can be defined as follows: 1) Distance: a. |XC|, (for example, K(X) = |XC|); b.|XC|>th*maxVal?0:X, where th represents the threshold. For example, th = 0.5. maxVal is 2 BitDepth It is defined as -1, minVal is 0, and midVal=2 (BitDepth-1) That is the case. Note: x?y:z indicates that if x is true or not equal to 0, the output is the value of y; otherwise, the output is the value of z. 2) Square: a. (X*|XC|+midVal)>>bitDepth, b.[(XC) 2 +midVal]>>bitDepth 3) Square root: a.sqrt(X*|XC|) 4) Gauss: aX*exp[-|XC| / maxVal], bX*exp[-(XC) 2 / maxVal 2 ] 5) Polynomial aX*(1-|XC| / maxVal), bX*(1-|XC| 2 / maxVal 2 ), cX*(1-|XC| 3 / maxVal 3 ), etc.

[0043] In another embodiment, a higher-order polynomial (e.g., cubic) of the central chroma sample can be used to replace the original nonlinear term P=(C*C+midVal)>>bitDepth. In that case, the predicted chroma value can be defined as follows: predChromaVal =c0C+c1N+c2S+C3E+c4W+c5T+c6B (10) Here, the nonlinear term is T=(C 3 (+midVal * maxVal) >> 2 * bit depth.

[0044] In another embodiment, while preserving the nonlinear term P=(C*C+midVal)>>bitDepth, a higher-order polynomial (e.g., cubic) of the central lumern sample can be used to replace one of the four neighboring samples furthest from the central lumern sample. The predicted chroma value can be defined as follows:

number

[0045] In another embodiment, instead of replacing the original 7-tap filter, the modified NL-CCCM filter described above can also be treated as an additional intra-prediction mode. Rate distortion (RD) determination on the encoder side is based on the RD cost. When CCCM is enabled, a CU level flag is signaled to indicate whether the additional NL-CCCM mode should be used. If the NL-CCCM mode is used, the following filter indices are signaled to indicate which nonlinear filter is applied to derive the final chroma prediction value. An example of the syntax nl_cccm_flag may be used. A nl_cccm_flag equal to 1 indicates that nonlinear CCCM will be used. A nl_cccm_flag equal to 0 indicates that nonlinear CCCM will not be used.

[0046] [References] Each of the references listed here is referenced in full. The term JVET refers to Joint Video Experts as defined in ITU-T SG 16 WP 3 and ISO / IEC JTC 1 / SC 29. [1] “Versatile Video Coding”, Rec. ITU-T H.266, August 2020 [2]JVET-AB2025, “Algorithm description of Enhanced Compression Model 7 (ECM 7)”, M. Coban et al., Mainz, Germany, October 2022

[0047] [Example of computer system implementation] Embodiments of the present invention may be implemented by computer systems, systems composed of electronic circuits and components, integrated circuit (IC) devices such as microcontrollers, field-programmable gate arrays (FPGAs), other configurable or programmable logic devices (PLDs), discrete-time or digital signal processors (DSPs), application-specific ICs (ASICs), and / or devices including one or more such systems, devices, or components. Computers and / or ICs may implement, control, or execute instructions relating to the application of cross-component prediction tools in image and video coding, such as those described herein. Computers and / or ICs may calculate any of the various parameters or values ​​relating to the application of cross-component prediction tools in image and video coding, as described herein. Embodiments of images and videos may be implemented in hardware, software, firmware, and various combinations thereof.

[0048] A particular embodiment of the present invention includes a computer processor that executes software instructions, which cause the processor to perform the method of the present invention. For example, one or more processors, such as a display, encoder, set-top box, or transcoder, may implement the method relating to the application of the cross-component prediction tool in image and video coding described above by executing software instructions in program memory accessible to the processor. Embodiments of the present invention may also be provided in the form of a program product. A program product may include any non-temporary and tangible case that, when executed by a data processor, carries a set of computer-readable signals containing instructions that cause the data processor to perform the method of the present invention. A program product according to the present invention may take any of a wide variety of non-temporary and tangible forms. A program product may include, for example, a physical medium such as a magnetic data storage medium such as a floppy diskette or hard disk drive, an optical data storage medium such as a CD-ROM or DVD, or an electronic data storage medium such as ROM or flash RAM. The computer-readable signals on the program product may optionally be compressed or encrypted.

[0049] Where components (e.g., software modules, processors, assemblies, devices, circuits, etc.) are referenced above, unless otherwise indicated, references to such components (including references to “means”) should be interpreted as including any components that perform the function of the described component (e.g., functionally equivalent) (including components that perform the function in the described embodiments of the invention but are not structurally equivalent to the disclosed structures).

[0050] Various aspects of this disclosure can be understood from the following Enumerated Example Embodiments (EEE): EEE1: A method for video decoding an intercoded coding unit, Receiving coding units (CUs) coded in interpredictive mode, The first flag detects whether chroma prediction is enabled by linear model luma-dependent chroma interpretation, If the first flag indicates that chroma prediction is enabled by linear model luma-dependent prediction, Read the first syntax parameter that indicates a specific ruma-dependent chroma-inter prediction mode among two or more ruma-dependent chroma-inter prediction modes, Read the second flag indicating whether or not a convolutional cross-component (CCCM) interpredictive model is being used. Decode the chroma pixels of the coding unit based on the specified chroma-dependent chrominator prediction mode and the second flag. A method of having. The EEE2.EEE1 method, The first syntax parameter indicates one of either a cross-component linear model (CCLM) or a gradient linear model (GLM) used for chroma prediction of the intracoded CU. method. The EEE3.EEE2 method, The above first syntax parameter indicates the use of a fusion of the CCLM model and the CCCM model. method. The EEE4.EEE3 method, The aforementioned fusion is T Ref =w*T Inter +(1-w)*T LMC It can be expressed as, T Ref This represents the reference template region of the chroma reconstruction strip adjacent to the coding unit, T InterThis represents a chroma prediction sample derived using interpretation for the aforementioned reference template region. T LMC This represents a chroma prediction sample derived using LMC prediction for the aforementioned reference template region. w represents the weight within [0,1]. method. EEE5. A method for encoding a video bitstream using cross-component residual modeling (CCRM), Each frame receives an input video picture containing lumen pixels and chroma pixels, Regarding the area within the input video picture that should be coded, Access the original chroma pixel values ​​of the region, Access the predicted chroma pixel values ​​and predicted residual luma pixel values ​​for the region, A chroma residual value is generated by subtracting the predicted chroma pixel value from the corresponding original chroma pixel value. Based on the predicted residual lumen pixel values ​​and chroma residual values, the CCRM model parameters are derived. The CCRM model parameters are applied to the predicted residual lumen pixel values ​​to generate adjusted predicted residual lumen pixel values. The adjusted predicted residual lumen pixel value is subtracted from the chroma residual value to generate the adjusted chroma residual value. To generate a coded bitstream based at least on the adjusted chroma residual values. A method of having. The EEE6.EEE5 method, The CCRM model parameters include the alpha value, Calculating the aforementioned alpha value means

number

number

number

[0051] [Equivalent, extension, substitute, other] Exemplary embodiments relating to the application of cross-component predictive coding tools in image and video coding have been described herein. Embodiments of the present invention have been described herein with reference to numerous specific details, which may vary from one embodiment to the next. Thus, the sole and exclusive indicator of what the present invention is and what the applicant intends it to be is the set of claims arising from this application, including the specific form in which such claims were issued and any subsequent modifications. The definitions expressly provided herein for terms contained in such claims shall define the meaning of the terms used in such claims. Accordingly, no limitations, elements, characteristics, features, advantages, or attributes not expressly stated in a claim shall in any way limit the scope of such claim. Accordingly, the specification and drawings should be interpreted as illustrative and not restrictive.

[0052] [Cross-reference of related applications] This application claims priority from Indian Provisional Patent Application No. 202311027178, filed on 12 April 2023, and Indian Provisional Patent Application No. 202311038996, filed on 7 June 2023, each of which is incorporated herein by reference in its entirety.

Claims

1. A method for video decoding an intercoded coding unit, Receiving coding units coded in interprediction mode, The first flag detects whether chroma prediction is enabled by linear model luma-dependent chroma interpretation, If the first flag indicates that chroma prediction is enabled by linear model luma-dependent prediction, Read the first syntax parameter that indicates a specific ruma-dependent chroma-inter prediction mode among two or more ruma-dependent chroma-inter prediction modes, Read the second flag indicating whether a convolutional cross-component (CCCM) interpredictive model is being used. Decode the chroma pixels of the coding unit based on the specified chroma-dependent chrominator prediction mode and the second flag. A method of having.

2. The first syntax parameter indicates one of either a cross-component linear model (CCLM) or a gradient linear model (GLM) used for chroma prediction of the intracoded CU. The method according to claim 1.

3. The above first syntax parameter indicates the use of a fusion of the CCLM model and the CCCM model. The method according to claim 2.

4. The aforementioned fusion is T Ｒｅｆ =w*T Ｉｎｔｅｒ +(1-w)*T ＬＭＣ It can be expressed as, T Ｒｅｆ This represents the reference template region of the chroma reconstruction strip adjacent to the coding unit, T Ｉｎｔｅｒ This represents a chroma prediction sample derived using interpretation for the aforementioned reference template region. T ＬＭＣ This represents a chroma prediction sample derived using LMC prediction for the aforementioned reference template region. w represents the weight within the range [0, 1]. The method according to claim 3.

5. A method for encoding a video bitstream using cross-component residual modeling (CCRM), Each frame receives an input video picture containing lumen pixels and chroma pixels, Regarding the area within the input video picture that should be coded, Access the original chroma pixel values ​​of the region, Access the predicted chroma pixel values ​​and predicted residual luma pixel values ​​for the region, A chroma residual value is generated by subtracting the predicted chroma pixel value from the corresponding original chroma pixel value. Based on the predicted residual lumen pixel values ​​and chroma residual values, the CCRM model parameters are derived. The CCRM model parameters are applied to the predicted residual lumen pixel values ​​to generate adjusted predicted residual lumen pixel values. The adjusted predicted residual lumen pixel value is subtracted from the chroma residual value to generate the adjusted chroma residual value. To generate a coded bitstream based at least on the adjusted chroma residual values. A method of having.

6. The CCRM model parameters include alpha values. Calculating the aforementioned alpha value means [Math 1] This includes calculating Rees Ｃ This represents the predicted chroma residual value, and Res' Ｌ represents the downsampled predicted lumer residual value around the region, and N represents the total number of neighboring samples. The method according to claim 5.

7. The CCRM model parameters are signaled to the decoder along with the coded bitstream. The method according to claim 5 or 6.

8. A method for decoding a coded video bitstream using cross-component residual modeling (CCRM), The system receives a coded bitstream containing coded pictures, each coded picture containing lumen pixels and chroma pixels, Regarding the area within the coded video picture, Based on the coded bitstream, the predicted residual lumen pixel values ​​and predicted residual chroma pixel values ​​for the region are generated. Based on the predicted residual lumen pixel values ​​and the predicted residual chroma pixel values, access the CCRM model parameters. The CCRM model parameters are applied to the predicted residual lumen pixel values ​​to generate adjusted predicted residual lumen pixel values. Adding the adjusted predicted residual lumen pixel value to the predicted residual chroma pixel value, an adjusted chroma residual value is generated. The adjusted chroma residuals are added to the interpredicted chroma values ​​for the region to generate the output chroma pixel values ​​for that region. A method of having.

9. The CCRM model parameters include alpha values. Calculating the aforementioned alpha value means [Math 2] including calculating Res Ｃ represents the predicted chroma residual value, and Res’ Ｌ represents the downsampled predicted luma residual value around the region, and N represents the total number of neighboring samples The method according to claim 8.

10. The CCRM model parameters are received along with the coded bitstream. The method according to claim 8 or 9.

11. The CCRM model parameters further include a CCRM merge flag to indicate whether CCRM merging is enabled or not. When CCRM is enabled, When CCRM merging is enabled, CCRM model parameters are inferred from neighboring CUs. If CCRM merging is not enabled, the CCRM model parameters are extracted from the bitstream. The method according to any one of claims 8 to 10.

12. The CCRM model parameter includes an index to an array of possible alpha values ​​and a sign flag indicating whether the alpha value selected by the index is positive or negative. The method according to any one of claims 8 to 11.

13. A method for chroma prediction using a convolutional cross-component (CCCM) prediction model, The aforementioned model, predChromaVal =c ０ C+c １ K(N)+c ２ K(S)+c ３ K(E)+c ４ K(W)+c ５ P+c ６ B This includes predicting the chroma value using a nonlinear filter according to c ｉ Here, for i = 0 to 6, the filter parameters are represented, K(.) is the nonlinear kernel function, and for the central lumer sample (C), N, S, W, and E represent the neighboring pixel samples to the north, south, west, and east, respectively, and P = (C * C + midVal) >> bitDepth, where midVal = 2 （ｂｉｔＤｅｐｔｈ－１） And B is from 0 to 2 ｂｉｔＤｅｐｔｈ This is a fixed bias term between -1, and for neighboring pixels X, K(X) is: K(X) = |X - C| or K(X)=|X−C|>th*maxVal? 0:X It has a distance function such that th represents the threshold, and maxVal = 2 ｂｉｔＤｅｐｔｈ -1 method.

14. K(X) is, K(X)=(X*|X-C|+midVal) >>bitDepth, (()) ２ __________________________________________________________ K(X)=sqrt(X*|X−C|), K(X)=X*exp[-|X-C| / maxVal], K(X)=X*exp[-(X-C) ２ / maxVal ２ ]、 K(X)=X*(1-|X-C| / maxVal), K(X)=X*(1-|X-C| ２ / maxVal ２ ), or K(X)=X*(1-|X-C| ３ / maxVal ３ ) Including one of the following, The method according to claim 13.

15. A method for chroma prediction using a convolutional cross-component (CCCM) prediction model, The aforementioned model, predChromaVal =c ０ C+c １ N+c ２ S+c ３ E+c ４ W+c ５ T+c ６ B This includes predicting the chroma value using a nonlinear filter according to c ｉ For i = 0 to 6, represents the filter parameter, and for the central lumen sample (C), N, S, W, and E represent the neighboring pixel samples to its north, south, west, and east, respectively. T=(C ３ +2 （ｂｉｔＤｅｐｔｈ－１） *2 ｂｉｔＤｅｐｔｈ -1)>>2*bitDepth And B is from 0 to 2 ｂｉｔＤｅｐｔｈ The fixed bias term between -1 is method.

16. A method for chroma prediction using a convolutional cross-component (CCCM) prediction model, The aforementioned model, [Math 3] This includes predicting the chroma value using a nonlinear filter according to c ｉ Here, when i = 0 to 6, the filter parameter is represented, and B is from 0 to 2 ｂｉｔＤｅｐｔｈ This is a fixed bias term between -1, For the central luma sample (C), X ｉ This represents one of its east (E), north (N), west (W), and south (S) neighborhoods, and the neighborhood whose lumen value differs most from the lumen value of the central lumen sample (C) is excluded. T=(C ３ +2 （ｂｉｔＤｅｐｔｈ－１） *2 ｂｉｔＤｅｐｔｈ -1)>>2*bitDepth That is, method.

17. A tangible computer-readable storage medium storing computer-executable instructions for one or more processors to perform the method according to any one of claims 1 to 16.

18. An apparatus comprising a processor and configured to perform the method described in any one of claims 1 to 16.

Images