Inter prediction and intra prediction method using sample-side filter, video coding device and computer readable medium

Abstract

This application provides an intra-frame prediction method and a color component prediction method. The intra-frame prediction method includes: obtaining color component samples from a template region; applying a first sample-side filter (SSF) to the color component samples; obtaining a linear function based on the filtered color component samples of the template region, the linear function being used to predict the chrominance component based on the luminance component; obtaining a luminance sample of the current block; and obtaining a chrominance prediction of the current block based on the linear function and the luminance sample of the current block. Implementation of these embodiments can improve data compression efficiency.

Description

Technical Field

[0001] This application relates to the field of encoding and decoding technology, specifically to an intra-frame prediction method based on sample-side filters, an intra-frame or inter-frame prediction method, a video encoding and decoding device, and a computer-readable medium. Background Technology

[0002] In video compression, a color image or a frame of color video typically consists of three color components: a luminance component (Y) and two chrominance components (Cb and Cr). Each component is represented as a data matrix, and this matrix is ​​divided into blocks associated with specific coding parameters. Any block for a component can be a square or a rectangle with a side length that is a power of 2. Image encoding is performed in raster scan order: first from left to right, then from top to bottom, and within a specific block or blocks, the luminance component is processed first, followed by the chrominance components.

[0003] Popular video coding standards such as Versatile Video Coding (VVC) use a hybrid prediction or transform coding framework. Prediction refers to predicting the current block (the block to be encoded) using coded blocks or regions within the same frame (i.e., intra-frame prediction) or coded blocks or regions from different frames (i.e., inter-frame prediction). When performing intra-frame or inter-frame prediction, the encoder tries multiple intra-frame prediction modes that conform to the coding standard, calculates and compares the corresponding predicted blocks, and selects the best prediction mode. The difference between the current block and the best predicted block, i.e., the residual, is also encoded. By transmitting only the prediction mode and the residual, the encoder can instruct the decoder to decode and reconstruct the original image or video and its approximation.

[0004] In VVC and recent research on future video coding standards, some prediction models use linear models to generate prediction blocks, including the Cross-Component Linear Model (CCLM), Gradient Linear Model (GLM), Convolutional Cross-Component Model (CCCM), and Local Illumination Compensation (LIC).

[0005] The prediction models described above share commonalities. They generally involve establishing a linear model, solving a linear regression problem, and applying this model to predict the image. First, a formula for generating prediction blocks can be defined, which is a linear function of the reconstructed samples and known attributes. Then, the coefficients in the formula can be derived using samples from the template region relative to the reference block and the current block. For example, in CCLM, a chroma block is predicted from a luminance block at the same location, using samples of chroma values ​​and luminance values ​​at the same location from the template region. The derivation of the coefficients is the process of solving a linear regression problem using these samples, and the derived formula can then be applied to the current block to generate prediction blocks.

[0006] However, linear regression is highly sensitive to noise, especially with small sample sizes. Because linear regression uses the L2 norm between predicted and sample values ​​as its minimization objective, the solution model may be significantly biased towards sample points that deviate from a linear model, thus increasing the prediction error for most samples. This is particularly pronounced when there are large outliers in the input data.

[0007] Furthermore, linear regression using least squares provides the best estimate in terms of squared error. It exhibits excellent predictive performance when the data is stable and contains Gaussian noise, but its generalization ability is insufficient. Predictive performance deteriorates significantly when the model for the test sample differs from the model in the training set, or when non-Gaussian noise is present. Since block boundaries in video coding are pre-defined grids rather than natural object boundaries, the aforementioned factors frequently arise. For these reasons, incorporating models that resist outliers is a more preferable choice. Summary of the Invention

[0008] To address the aforementioned problems, this application provides an intra-frame prediction method with a sample-side filter, an intra-frame or inter-frame prediction method, a video encoding / decoding device, and a computer-readable storage medium.

[0009] One aspect of this application provides an intra-frame prediction method, which includes: obtaining color component samples from a template region; applying a first sample-side filter (SSF) to the color component samples; obtaining a linear function based on the filtered color component samples of the template region; obtaining a luminance sample of the current block; and generating a chromaticity prediction of the current block based on the linear function and the luminance sample of the current block.

[0010] Another aspect of this application provides an intra-frame prediction method. The method includes: acquiring color component samples from a template region; obtaining a linear function for predicting chromaticity components based on the color component samples from the template region; acquiring a luminance sample of the current block; applying a second SSF to the luminance sample of the current block; and obtaining a chromaticity prediction of the current block based on the linear function and the filtered luminance sample of the current block.

[0011] Another aspect of this application provides an intra-frame prediction method. The method includes: acquiring color component samples from a template region; applying a first SSF to the color component samples from the template region; obtaining a linear function for predicting chrominance components based on the filtered color component samples from the template region; acquiring luminance samples of the current block; applying a second SSF to the luminance samples of the current block; obtaining a chrominance prediction for the current block based on the linear function for predicting chrominance components based on the luminance components and the filtered luminance samples of the current block; and applying a third SSF to the chrominance prediction for the current block.

[0012] Another aspect of this application provides an intra-frame or inter-frame prediction method. The method includes: configuring a sample-side filter (SSF); determining one or more specific samples, said specific samples being from at least one template region or current block; applying the SSF to the one or more specific samples; and performing color component prediction based on the filtered specific samples.

[0013] This application also provides a video encoding / decoding apparatus. The apparatus includes a processor and a memory. The memory is configured to store executable instructions that, when executed by the processor, cause the processor to perform any of the aforementioned methods.

[0014] This application also provides a computer-readable medium storing executable instructions that, when executed by a processor, cause the processor to perform any of the aforementioned methods.

[0015] The method provided in this application embodiment is to use filters for linear prediction mode samples during image and video encoding to perform intra-frame or inter-frame prediction, which can improve data compression efficiency. Attached Figure Description

[0016] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments of this application and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a schematic diagram of a video encoding system based on related technologies according to an embodiment of this application; Figure 2This is a schematic diagram of a video decoding system based on related technologies according to an embodiment of this application; Figure 3 This is a schematic diagram illustrating the definition of a template region in a color component prediction method according to an embodiment of this application; Figure 4 This is a flowchart illustrating a CCLM mode decoding process based on related technologies, as described in an embodiment of this application. Figure 5 This is a flowchart illustrating a LIC mode decoding process based on related technologies, as described in an embodiment of this application. Figure 6 This is a flowchart of an intra-frame prediction method according to an embodiment of this application; Figure 7 This is a flowchart of another intra-frame prediction method involved in the embodiments of this application; Figure 8 This is a flowchart of another intra-frame prediction method involved in the embodiments of this application; Figure 9 This is a flowchart of an intra-frame or inter-frame prediction method according to an embodiment of this application; Figure 10 This is a flowchart of an intra-frame prediction method in CCCM mode according to an embodiment of this application; Figure 11 This is a schematic diagram illustrating a sample-side filter applied to a luminance or chromaticity sample in a template region, according to an embodiment of this application. Figure 12 This is an exemplary Sobel operator for a luminance sample in a GLM mode, as described in an embodiment of this application. Figure 13 This is a schematic diagram of the pixel to be processed and its predicted neighboring pixels in an extrapolation filter-based intra prediction (EIP) process according to an embodiment of this application. Figure 14 This is a schematic diagram of an encoder or decoder according to an embodiment of this application. Detailed Implementation

[0017] The technical solutions of the embodiments of this application will now be described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0018] To help understand the technical solutions proposed in the various embodiments disclosed in this application, a video encoding and decoding system will be briefly introduced below.

[0019] Figure 1 A video coding system based on related technologies is shown, such as Figure 1 As shown, the video coding system 110 consists of multiple modules, including a block partitioning unit 1101, a transform and quantization unit 1102, an intra-frame estimation unit 1103, an intra-frame prediction unit 1104, a motion compensation unit 1105, a motion estimation unit 1106, an inverse transform and inverse quantization unit 1107, a filter control and analysis unit 1108, a filtering unit 1109, a coding unit 1110, a coded image buffer unit 1111, and a subtractor 1112.

[0020] The original video signal consists of video frames. Each video frame can be divided into video blocks by the block partitioning unit 1101. For each video frame, the subtractor 1112 generates residual pixel information of the residual frame by subtracting the prediction block output by the intra-frame prediction unit 1104 or the motion compensation unit 1105 from the video frame. The residual pixel information obtained after intra-frame prediction or inter-frame prediction (motion compensation) is transformed by the transform and quantization unit 1102, which includes transforming the residual pixel information from the pixel domain to the transform domain and quantizing the resulting transform coefficients to further reduce the bit rate. The intra-frame estimation unit 1103 performs intra-frame estimation, and the intra-frame prediction unit 1104 performs intra-frame prediction on the video reconstruction block. The motion estimation performed by the motion estimation unit 1106 generates motion vectors that can estimate the motion of the video reconstruction block, and then the motion compensation unit 1105 performs motion compensation based on the determined motion vectors. After determining the intra-prediction mode, the intra-prediction unit 1104 provides the selected intra-prediction data to the coding unit 1110, and the motion estimation unit 1106 also sends the calculated motion vector data to the coding unit 1110. The inverse transform and inverse quantization unit 1107 reconstructs the video reconstruction block and reconstructs the residual block in the pixel domain. The filtering unit 1109, controlled by the filtering control and analysis unit 1108, removes block artifacts from the reconstructed residual block, and the coding unit 1110 adds the reconstructed residual block to the prediction block of the encoded image buffer unit 1111 to generate the reconstructed block. The coding unit 1110 encodes various coding parameters and quantization transform coefficients into a bitstream and outputs the bitstream of the video signal. The encoded image buffer unit 1111 stores the reconstructed block as a reference block for intra-prediction. As video image coding progresses, new reconstructed blocks are continuously generated, and these blocks are stored in the encoded image buffer unit 1111.

[0021] Figure 2 A video decoding system based on related technologies is shown, such as Figure 2As shown, the video decoding system 120 consists of multiple modules, including a decoding unit 1201, an inverse transform and inverse quantization unit 1202, an intra-frame prediction unit 1203, a motion compensation unit 1204, a filtering unit 1205, a decoded image buffer unit 1206, and a post-filtering unit 1207.

[0022] The input signal of a video frame is encoded by a video encoding system 110 to obtain an output bitstream, which is then transmitted to a video decoding system 120. The video decoding system 120 receives the bitstream representing the video frame in an encoded format (i.e., a compressed format). In the video decoding system 120, the bitstream is processed by a decoding unit 1201 to obtain decoded transform coefficients. An inverse transform and inverse quantization unit 1202 processes these transform coefficients to generate residual blocks in the pixel domain. An intra-prediction unit 1203 generates intra-prediction blocks for the current video decoding block based on a determined intra-prediction mode and data from previous decoded blocks of the current video frame or image. A motion compensation unit 1204 determines the inter-prediction information for the current video decoding block and generates inter-prediction blocks by parsing motion vectors and other relevant syntax elements. Finally, the decoded video block is obtained by adding the residual block from the inverse transform and inverse quantization unit 1202 to the corresponding prediction block generated by the intra-prediction unit 1203 or the motion compensation unit 1204. To improve video quality, the decoded video block is filtered by the filtering unit 1205 to remove block artifacts. After filtering, the decoded video block is stored in the decoded image buffer unit 1206 as a reference block for subsequent intra-frame prediction or motion compensation, or for video output, i.e., to reproduce and reconstruct the original video signal. Optionally, the output video can be further processed by the post-filtering unit 1207 to obtain a better or enhanced viewing experience.

[0023] In recent research on VVC and future-oriented video coding standards, some prediction models use linear models to generate prediction blocks. Examples include the cross-component linear model CCLM, the gradient linear model GLM, the convolutional cross-component model CCCM, and the local illumination compensation model LIC.

[0024] The prediction models described above share commonalities. They generally involve establishing a linear model, solving a linear regression problem, and then applying this model to predict the image. First, a formula for generating prediction blocks can be defined, which is a linear function of the reconstructed samples and known attributes. Then, the coefficients in the formula can be derived using samples from the template region relative to the reference block and the current block. For example, in CCLM, a chroma block is predicted from a luminance block at the same location, using samples of chroma values ​​and luminance values ​​at the same location from the template region. The derivation of the coefficients is essentially the process of solving a linear regression problem using the aforementioned samples. The resulting formula can then be applied to the current block to generate prediction blocks.

[0025] The aforementioned predicted block that generates the current block The process can be expressed as a formula:

[0026]

[0027] In this formula, It is a linear prediction function. Through template area The optimal function derived from the samples. This is achieved by minimizing the template region of the current block. Reconstructed samples With the template area of ​​the reference block Reconstructed samples It is obtained by substituting the sum of the squared differences between the predicted values ​​generated by the prediction function. Using and reference block The reconstructed samples can generate predicted values ​​for the current block. Typically, the template region is defined as the left and top adjacent regions of the reference block and the current block, such as... Figure 3 As shown, the predictive power of the linear model comes from the spatial proximity between the block and the template.

[0028] In this application, these modes are generally referred to as linear model prediction modes, such as cross-component linear model (CCLM), gradient linear model (GLM), convolutional cross-component model (CCCM), intra-frame prediction based on extrapolation filter (EIP), local illumination compensation (LIC), multi-model linear model (MMLM), and local-boosting cross-component prediction (LBCCP).

[0029] Figure 4 A flowchart of a CCLM mode decoding process based on related technologies is shown, such as... Figure 4 As shown, the first step is to determine whether CCLM mode is enabled. If CCLM mode is not enabled, other prediction modes are decoded; if CCLM mode is enabled, luminance and chrominance samples from the template region are obtained to derive the luminance-chrominance linear formula, and then the chrominance prediction block can be calculated based on the formula.

[0030] Figure 5 A flowchart illustrating a LIC mode decoding process based on related technologies is shown. Figure 5As shown, the first step is to determine whether LIC mode is enabled. If LIC mode is not enabled, other prediction modes are decoded. If LIC mode is enabled, reconstructed samples from templates of the reference block and the current block are obtained to obtain a linear formula for the reference block versus the current block, which is then used to calculate the predicted block.

[0031] One objective of this application is to mitigate the distortion of linear regression results caused by noise in the reconstructed samples. To this end, this application introduces a sample-side filter (SSF), primarily involving the application of the sample-side filter to linear model prediction. Linear model prediction refers to any intra-frame or inter-frame prediction mode based on a linear model to generate a predicted image, the coefficients of which are obtained by solving the same linear regression problem on both the encoder and decoder sides. The SSF is a filter used to generate the predicted image on top of the input samples and / or their variants used to perform linear regression.

[0032] Figure 6 A flowchart of an intra-frame prediction method is shown, such as... Figure 6 As shown, the method includes: obtaining color component samples from a template region (S201); applying a first sample-side filter (SSF) to the color component samples within the template region (S202); obtaining a linear function for predicting chromaticity components from luminance components based on the filtered color component samples of the template region (S203); obtaining luminance samples of the current block (S204); and generating a chromaticity prediction for the current block based on the linear function for predicting chromaticity components from luminance components and the luminance samples of the current block (S205).

[0033] In S201, color component samples are obtained from the template region.

[0034] Color component samples include luminance samples and / or chrominance samples, and the template region can be determined based on the actual prediction mode enabled for the intra-frame prediction process.

[0035] In S202, the first sample-side filter (SSF) is applied to the color component samples within the template region.

[0036] The first SSF can be designed as a low-pass filter in the spatial domain of an image or video frame. For ease of implementation, the SSF can be designed as a finite impulse response (FIR) filter, which operates only within the adjacent range of the pixel to be computed and within the same frame.

[0037] In one example, the SSF can be designed as a convolution between luminance samples and a filter kernel:

[0038]

[0039] in This represents a linear convolution operation. This represents the luminance component.

[0040] This operation can also be represented in the form of pixel-by-pixel processing:

[0041] in This indicates a right shift operation.

[0042] Optionally or additionally, SSF can be applied to chromaticity samples:

[0043] In another example, the coefficients of SSF can be changed relative to the previous example while retaining its "low-pass" and "FIR" characteristics:

[0044]

[0045] Similarly, this operation can also be represented in the form of pixel-by-pixel processing:

[0046] In another example, the SSF could be a 3x3 median filter, and also a low-pass and FIR filter:

[0047]

[0048] The output at that point is The input value is the median of the input values ​​of its 8 connected neighbors.

[0049] In another example, the size of the SSF can be reduced to 3x1 and 1x3, which can be used in CCLM mode, where the upper stencil region consists of only one row of pixels and the left stencil region consists of only one column of pixels:

[0050]

[0051] Another example applicable to cross-component prediction involves downsampling the luminance sample to match the size of the chrominance sample. In a typical YUV420 format, the width and height of the luminance component are twice that of the chrominance component:

[0052]

[0053] SSF can be integrated into downsampling operations, allowing downsampling and SSF to be completed in a single linear convolution computation:

[0054] In this operation, the first SSF is applied after sample collection and before obtaining the linear model. As previously mentioned, the color component samples described in S201 may include luminance samples and / or chromaticity samples. Accordingly, the first SSF can be applied to the reconstructed luminance samples in the template region:

[0055] Optionally or additionally, the first SSF can be applied to the reconstructed chromaticity sample in the template region:

[0056] In other words, filter and They can be used simultaneously, or only one of them can be used.

[0057] In S203, a linear function for predicting the chromaticity component from the luminance component is obtained based on the filtered color component samples of the template region.

[0058] Based on the aforementioned operations of filtering the color components of the wave, linear functions can be obtained in different ways: For example, if the luminance sample is filtered using SSF, a linear function for predicting the chrominance component based on the filtered luminance and chrominance samples can be established. If the chrominance sample is filtered using SSF, a linear function for predicting the chrominance component based on the luminance and filtered chrominance samples can be established. If both the luminance and chrominance samples are filtered using SSF, a linear function for predicting the chrominance component based on the filtered luminance and filtered chrominance samples can be established.

[0059] In S204, obtain the brightness sample of the current block.

[0060] The method for obtaining the brightness sample of the current block can be implemented using existing technologies, so it will not be described in detail here.

[0061] In S205, a chromaticity prediction for the current block is generated based on the linear function used to predict the chromaticity component from the luminance component and the luminance sample of the current block.

[0062] In summary, the method provided in this embodiment can apply a sample-side filter to color component samples from a template region. Implementation of this application helps mitigate the distortion of linear regression results caused by noise in the reconstructed samples, thus improving data compression efficiency.

[0063] In some embodiments, after obtaining the luminance sample of the current block, the method may further include: applying a second SSF to the luminance sample of the current block. Accordingly, the aforementioned operation of generating a chromaticity prediction of the current block based on the luminance sample of the current block using the linear function may further include: generating a chromaticity prediction of the current block based on the linear function and the filtered luminance sample of the current block.

[0064] The color component sample described in S202 can be a luminance sample from the template region. In this case, the first SSF applied to the luminance sample from the template region and the second SSF applied to the luminance sample of the current block can be the same. In other words, the same SSF used to filter the luminance sample for establishing the linear function for predicting the chromaticity component from the luminance component can be used as the same SSF used to generate the chromaticity prediction of the current block, thereby ensuring the accuracy of the prediction result.

[0065] In some embodiments, the method may further include: using a third SSF for chroma prediction of the current block. That is, the chroma prediction obtained in S205 can be further filtered using a third SSF to improve prediction accuracy. The third SSF may be the same as or different from the first SSF.

[0066] In some embodiments, the method may further include: determining whether a first SSF is available. Applying the first SSF to color component samples from a template region may include: in response to determining that the first SSF is available, applying the first SSF to the color component samples from the template region. The availability determination of the first SSF may be based on at least one of the following: one or more predefined prediction modes; the size of the block in the template region; or an indication received from the encoder indicating whether the first SSF is available.

[0067] In some embodiments, the method may further include: determining whether a first SSF is enabled. Whether the first SSF is enabled may be determined based on at least one of the following: receiving an indication from the encoder indicating whether the first SSF is enabled; or deducing it based on the decoder of the reconstructed image.

[0068] Figure 7 A flowchart of another intra-frame prediction method is shown, such as Figure 7 As shown, the method may include the operations described in S301 to S305: In S301, color component samples from the template region are obtained.

[0069] In some embodiments, the method may further include applying a first SSF to the color component samples from the template region after the operation of acquiring the color component samples from the template region. The definition of the first SSF can be found in [reference needed]. Figure 6 The method described herein will not be repeated here. Filtered color component samples from the template region can be used to subsequently obtain a linear function for predicting the chromaticity component based on the luminance component.

[0070] In S302, a linear function for predicting the chromaticity component based on the color component samples from the template region is obtained.

[0071] In S303, obtain the brightness sample of the current block.

[0072] In S304, the second SSF is used for the luminance sample of the current block.

[0073] The second SSF can be designed as a low-pass filter in the spatial domain of an image or video frame. For ease of implementation, the SSF can be designed as an FIR filter, which operates only within the adjacent range of the pixel to be computed and within the same frame.

[0074] In other words, SSF can be applied before generating the predicted image. For example, in CCCM or CCLM mode, SSF can be applied to reconstruct luma blocks:

[0075] In S305, the chromaticity prediction of the current block is obtained based on the linear function used to predict the chromaticity component from the luminance component and the filtered luminance sample of the current block.

[0076] Accordingly, the filtered luminance samples can be used to generate chromaticity prediction blocks. In the aforementioned CCCM or CCLM examples, the filtered luminance samples... Used to generate chromaticity prediction blocks. The prediction process can be expressed by the formula:

[0077]

[0078] In some embodiments, the method may further include: applying a third SSF to the chroma prediction of the current block. In the aforementioned CCCM or CCLM examples, the SSF may also be applied to the chroma prediction block to obtain a new chroma prediction block.

[0079] It should be understood that the first SSF, the second SSF, and the third SSF can be the same or different from each other.

[0080] In summary, the method provided in this embodiment can apply a sample-side filter to the brightness sample of the current block. This helps to reduce the distortion of linear regression results caused by noise in the reconstructed samples, thus improving data compression efficiency.

[0081] Figure 8 A flowchart of an intra-frame prediction method is shown according to another embodiment of this application, such as... Figure 8 As shown, the method may include the operations described in S401 to S407: In S401, color component samples are obtained from the template region.

[0082] In S402, the first SSF is used for the color component sample from the template region.

[0083] In S403, a linear function for predicting the chromaticity component based on the filtered color component samples from the template region is obtained.

[0084] In S404, obtain the brightness sample of the current block.

[0085] In S405, the second SSF is used for the luminance sample of the current block.

[0086] In S406, the chromaticity prediction of the current block is obtained based on the linear function used to predict the chromaticity component from the luminance component and the filtered luminance sample of the current block.

[0087] In S407, the third SSF is used for chromaticity prediction of the current block.

[0088] In this embodiment, a sample-side filter can be enabled for linear regression and calculating the predicted image. The prediction process can be expressed as a formula:

[0089]

[0090]

[0091]

[0092]

[0093]

[0094] The explanation of the aforementioned formula has been given in the previous embodiments, and therefore will not be repeated here. The above prediction process is applicable at least to CCCM and CCLM prediction models.

[0095] The first SSF, the second SSF, and the third SSF can be low-pass filters and FIR filters. Furthermore, the first SSF, the second SSF, and the third SSF can be the same or different from each other.

[0096] The operation of applying a first SSF to a color component sample from a template region includes at least one of the following: applying the first SSF to a luminance sample from a template region; or applying the first SSF to a chromaticity sample from a template region. In some embodiments, the first SSF may be applied to a luminance sample from a template region, and correspondingly, the first SSF and the second SSF are the same.

[0097] In some embodiments, the method may further include: determining whether a first SSF, a second SSF, and a third SSF are available. The availability of the first SSF, the second SSF, and the third SSF may be determined based on at least one of the following: one or more predefined prediction modes; the size of the template region block; or receiving an indication of whether the first sample-side filter from the encoder is available.

[0098] In some embodiments, the method may further include: receiving an instruction for enabling at least one of a first SSF, a second SSF, or a third SSF.

[0099] In summary, the method provided in this embodiment can apply a sample-side filter to color component samples from a template region, apply a sample-side filter to the luminance sample of the current block, and apply a sample-side filter to the chromaticity prediction result. The method provided in this embodiment helps to reduce the distortion of linear regression results caused by noise in the reconstructed samples, thus improving data compression efficiency.

[0100] In some embodiments, the method is tested on the Enhanced Compression Model (ECM) software maintained by the ITU-T / ISO / IEC Joint Video Experts Team, which is available at: https: / / vcgit.hhi.fraunhofer.de / ecm / ECM.git. The aforementioned software version that can be used and tested is ECM-11.0, and its implementation details can be found at: M. Coban, R.-L. Liao, K. Naser, J. Ström, L. Zhang, "Algorithm description of Enhanced Compression Model 11 (ECM 11)", document JVET-AF2025, Oct. 2023 .

[0101] The test video sequence and parameter settings follow the requirements of the Common Test Conditions (CTC), details of which can be found here: M. Karczewicz, Y. Ye, "Common Test Conditions and evaluation procedures for enhanced compression tool testing”, document JVET- AF2017, Oct. 2023 .

[0102] For Class C, D, and E video sequences, the test results for the Bjøntegaard-Delta bitrate (BD-BR) of the Y, U, and V color channels under the "intra-frame" setting are shown in Table 1 below. Negative numbers indicate that bitrate savings were achieved at the same reconstruction quality, indicating that an improvement was achieved using this method.

[0103] Table 1 Test Results for Each Category of Test Sequences

[0104] Under the same testing conditions, this method was also implemented and tested on ECM-12.0, and detailed information can be found in the following literature: M. Coban, R.-L. Liao, K. Naser, J. Ström, L. Zhang. "Algorithm description of Enhanced Compression Model 12 (ECM 12)", document JVET-AF2025, January 2024 The results are shown in Table 2 below: Table 2 Test Results for Each Category of Test Sequence

[0105] The aforementioned test results demonstrate the beneficial effect of this method in improving video compression efficiency.

[0106] Figure 9 A flowchart of an intra-frame or inter-frame prediction method is shown, such as... Figure 9 As shown, the method may include the operations described in S501 to S504: In S501, a sample-side filter is provided.

[0107] The sample-side filter can be a low-pass filter or an FIR filter.

[0108] In S502, one or more specific samples are identified, wherein the specific samples are from at least one template region or the current block.

[0109] Specifically, you can first determine whether SSF is available.

[0110] Generally, SSF can be used for any linear model prediction mode, and its availability can be determined based on the following factors: • Prediction modes: SSF can be used for all prediction modes or prediction modes of non-empty subsets, including CCLM, GLM, all variants of CCCM, EIP and LIC and their variants; • Block size: SSF can be used for all block sizes or a specific block size; • High-level syntax: Send control flags at the slice level, image level, or image group level to indicate whether SSF is available.

[0111] In addition, it can be determined whether SSF is enabled and for which samples SSF is enabled based on at least one of the following: • Explicit indication: Signals indicate which SSFs are enabled; • Always enabled: SSF is enabled as long as it is available for the current block; • Obtained from the decoder: The decoder obtains whether SSF is enabled for the current block from the reconstructed image; • High-level syntax: Send control flags at the slice level, image level, or image group level to indicate the conditions under which SSF is enabled.

[0112] The following examples illustrate two explicit indication methods. In the first example, a flag is sent to indicate SSF is enabled in each of the following sample types: Table 3 Sample Types and Uses

[0113] In this example, the regression samples include, but are not limited to: luminance samples in inter-component prediction and reference block samples in local illumination compensation; the response samples include, but are not limited to: chromaticity samples in inter-component prediction and current block samples in local illumination compensation.

[0114] In the second example, a flag is sent to indicate whether SSF is enabled in each of the following sample types: Table 4 Sample Types and Uses

[0115] When an enabled SSF is applied to one or more samples, it is treated as a specific sample, and these specific samples are filtered using SSF in subsequent operations.

[0116] In S503, SSF is used for one or more specific samples.

[0117] It should be understood that in the presence of multiple specific samples, "SSF" may also include multiple SSFs. For example, if the specific samples include chroma samples from the template region and luminance samples from the current block, then the SSF may include a first SSF for filtering the chroma samples from the template region and a second SSF for filtering the luminance samples from the current block. The multiple SSFs may be the same or different, depending on the actual system design.

[0118] In S504, color component prediction is performed based on specific filtered samples.

[0119] The implementation methods under different prediction modes provided in the embodiments of this application are as follows: (1) Convolutional Cross-Component Model (CCCM) CCCM is an inter-component prediction model. CCCM differs from CCLM and GLM in that it typically has a larger template region and integrates various types of samples into the prediction formula.

[0120] The formula for predicting chromaticity blocks (Cb or Cr) in CCCM is as follows:

[0121] Where D is the color bit depth of the image.

[0122] Model coefficients in CCCM Obtained through the cost minimization formula:

[0123] Where R is the set of coordinates of all pixels in the template region, including the rows of pixels adjacent to the top of the current block and the columns of pixels adjacent to the left of the current block.

[0124] This numerical optimization problem can be solved using Gaussian elimination or other approximation algorithms. For detailed implementation information, please refer to: P. Astola, J. Lainema, RG Youvalari, A. Aminlou, K. Panusopone, EE2- 1.1a: Convolutional cross-component intra prediction model , document JVET-AA0126, Joint Video Experts Team (JVET), Jul. 2022.

[0125] Figure 10 A flowchart of an intra-frame prediction method for CCCM mode is shown. In this embodiment, sample-side filters can be used for model derivation and prediction. Specifically, the process includes: • Check if SSF is available and decode whether SSF is enabled for the current block; • Apply SSF to the luminance and / or chrominance samples in the reference template; • Calculate the prediction function coefficients of the convolutional cross-component model; • Apply SSF to the reconstructed brightness samples in the current block; • Generate a chroma prediction block for the current block; • Apply SSF to the chromaticity prediction block.

[0126] Figure 11 A schematic diagram is shown of a sample-side filter applied to a luminance or chromaticity sample in a template region.

[0127] In another embodiment for the CCCM mode, the sample-side filter is used only for the luminance component and is used for model derivation and prediction. Specifically, the process includes: • Check if SSF is available and decode whether SSF is enabled for the current block; • Apply SSF to the brightness samples in the reference template; • Calculate the prediction function coefficients of the convolutional cross-component model; • Apply SSF to the reconstructed brightness samples in the current block; Generate a chroma prediction block for the current block.

[0128] In another embodiment for CCCM mode, the sample-side filter may be used solely for model derivation. Specifically, the process includes: • Check if SSF is available and decode whether SSF is enabled for the current block; • Apply SSF to the luminance and / or chrominance samples in the reference template; • Calculate the prediction function coefficients of the convolutional cross-component model; Generate a chroma prediction block for the current block.

[0129] In another embodiment for CCCM mode, the sample-side filter may be used only for prediction. Specifically, the process includes: • Check if SSF is available and decode whether SSF is enabled for the current block; • Calculate the prediction function coefficients of the convolutional cross-component model; • Apply SSF to the reconstructed brightness samples in the current block; • Generate a chroma prediction block for the current block; • Apply SSF to the chromaticity prediction block.

[0130] (2) Multi-model linear model (MMLM) MMLM is an extension of linear model prediction. In recent research on video coding standards, MMLM is considered a variant of CCLM and CCCM, namely Multi-Model CCLM (MM-CCLM) and Multi-Model CCCM (MM-CCCM). Unlike ordinary linear model prediction methods, which treat the entire template region as a single training set and derive a linear model to predict the current block, MMLM divides all pixels in the template region into two training sets and derives a linear model for each set for prediction. For example, in MM-CCLM and MM-CCCM, pixels are classified based on whether the brightness value of a pixel in the template region is less than the average brightness value of the entire template region. Simultaneously, during the prediction phase, the same criterion can be used to classify pixels in the current block and apply the corresponding linear model for prediction.

[0131] The formula for predicting chromaticity blocks (Cb or Cr) in MM-CCLM is as follows:

[0132] In the above equation, It is the average value of the brightness samples in the template region. Function coefficients { , } is used to predict brightness values ​​less than or equal to pixels, and { , } Used to predict brightness values ​​greater than pixels, and These can be calculated separately:

[0133]

[0134]

[0135]

[0136] In the above equation, or It is the pixel value in the downsampled luminance template region. or It is the value of a pixel in the chroma stencil region. These are pixel coordinates. It corresponds to the sample in the template region. The average value; and Only includes brightness values ​​less than or equal to Pixels; and Only includes brightness values ​​greater than Pixels.

[0137] Similarly, in MM-CCCM, the formula for predicting chromaticity blocks (Cb or Cr) is as follows:

[0138] in,

[0139] Model coefficients in MM-CCCM and These can be calculated separately:

[0140]

[0141] In one embodiment for the MM-CCCM mode, a sample-side filter may be applied prior to pixel classification. Specifically, the process includes: • Check if SSF is available and decode whether SSF is enabled for the current block; • Apply SSF to the luminance and / or chrominance samples in the reference template; • Pixels are divided into two categories based on the average brightness value in the reference template; • Calculate the prediction function coefficients of the multi-model convolutional cross-component model; • Generate a chroma prediction block for the current block; • Apply SSF to the chromaticity prediction block.

[0142] In another embodiment for the MM-CCCM mode, a sample-side filter can be applied after pixel classification. Specifically, the process includes: • Check if SSF is available and decode whether SSF is enabled for the current block; • Pixels are divided into two categories based on the average brightness value in the reference template; • Apply SSF to the luminance and / or chrominance samples in each of the above categories; • Calculate the prediction function coefficients of the multi-model convolutional cross-component model; • Generate a chroma prediction block for the current block; • Apply SSF to the chromaticity prediction block.

[0143] (3) Gradient Linear Model (GLM) GLM is another inter-component prediction mode similar to CCLM. GLM does not directly use the luminance sample values ​​from CCLM, but instead uses the luminance sample gradients to derive the linear model. Therefore, the formula for predicting chromaticity blocks (Cb or Cr) is:

[0144] By applying such to brightness samples Figure 12 One of the Sobel operators shown is used to compute the gradient G. Details can be found at: 1) C.-W. Kuo, X. Xiu, N. Yan, H.-J. Zhu, W. Chen, H. Gao, X. Wan, AHG12: Enhanced CCLM, document JVET-Z0140, Joint Video Experts Team (JVET), April 2022 2) P. Astola, J. Lainema, RG Youvalari, A. Aminlou, K. Panusopone, C.-W. Kuo, H.-J. Zhu, X. Xiu, N. Yan, W. Chen, X. Wang, EE2-1.1c, 1.3a and 1.3b: Combined tests of EE2-1.1a, 1.1b and 1.2, document JVET- AA0126, Joint Video Experts Team (JVET), Jul. 2022 .

[0145] In one embodiment for implementing a sample-side filter in GLM mode, the process includes: • Check if SSF is available and decode whether SSF is enabled for the current block; • Apply SSF to the luminance and / or chrominance samples in the reference template; • Calculate the coefficients of the prediction function of the gradient linear model; • Apply SSF to the reconstructed brightness samples in the current block; • Generate a chroma prediction block for the current block; • Apply SSF to the chromaticity prediction block.

[0146] It should be understood that for GLM mode, SSF is not necessarily applied to all of the above samples. Exemplarily, in some embodiments, the SSF for GLM mode may be applied only to the luminance and chrominance samples in the reference template. In some embodiments, the SSF for GLM mode may be applied only to the reconstructed luminance samples in the current block.

[0147] (4) Cross-component linear model (CCLM) CCLM is a cross-component intra-prediction mode that uses a reconstructed luminance image or luminance patch to predict a chrominance image or chrominance patch. The term "cross-component" comes from the fact that prediction is performed from one color component to another.

[0148] The formula for predicting chromaticity blocks (Cb or Cr) in CCLM is as follows:

[0149] In the previous formula, It is the value of the pixel to be predicted in the chroma block. This is the value of the corresponding pixel in the reconstructed luminance block, which has been adjusted to the size of the chrominance block, and... These are pixel coordinates. Because in the 4:2:0 image format, the size (height and width) of the luminance image is twice the size of the chrominance image, it is generally done by reconstructing the luminance image... To obtain by downsampling .

[0150] Model coefficients and It is calculated based on the relationship between the luminance and chrominance images in the template region, which includes a row of pixels adjacent to the top of the current block and a column of pixels adjacent to the left of the current block:

[0151]

[0152] in, It is the pixel value in the downsampled brightness template region. It is the value of a pixel in the chroma stencil region. These are pixel coordinates. The sample in the corresponding template area The average value.

[0153] In the above implementation, the chroma block is the current block to be predicted, the luminance block is used as a reference block, and the template area is the reconstructed pixels on the left and top of each block.

[0154] In one embodiment for implementing a sample-side filter in CCLM mode, the process may include: • Check if SSF is available and decode whether SSF is enabled for the current block; • Apply SSF to the luminance and / or chrominance samples in the reference template; • Calculate the prediction function coefficients of the cross-component linear model; • Apply SSF to the reconstructed brightness samples in the current block; • Generate a chroma prediction block for the current block; • Apply SSF to the chromaticity prediction block.

[0155] It should be understood that for CCLM mode, SSF may not be applied to all of the above samples. Exemplarily, in some embodiments, SSF for CCLM mode may be applied only to the luminance and chrominance samples in the reference template; in some other embodiments, SSF for CCLM mode may be applied only to the reconstructed luminance samples in the current block.

[0156] (5) Local Illumination Compensation (LIC) LIC is a prediction enhancement method applied to inter-frame prediction. In inter-frame prediction, a reference block is found using the aforementioned motion compensation unit. To compensate for the illumination difference between the reference block and the current block, LIC is used to measure this difference and adjust the prediction block. The formula for generating the prediction block is:

[0157] in, It is the value of the pixel to be predicted. It is the value of the pixel in the reference block.

[0158] To calculate the model coefficients and A template region is defined around the reference block and the current block by selecting one or more rows adjacent to the top of each block and one or more columns adjacent to the left of each block. The calculation method is similar to that in CCLM described above.

[0159] In local illumination compensation, SSF can be applied to the reconstructed values ​​of the template region of the reference block and / or the template region of the current block. Filtered samples are used to derive the relevant linear model:

[0160]

[0161] In addition, SSF can also be applied to reference blocks:

[0162] The predicted value for the current block is:

[0163]

[0164] SSF can also be applied to prediction blocks to generate new prediction blocks:

[0165] In one embodiment for implementing a sample-side filter in LIC mode, the process may include: • Check if SSF is available and decode whether SSF is enabled for the current block; • Apply SSF to the reconstructed samples in the template region of the reference block and / or the current block; • Calculate the prediction function coefficients for local illumination compensation; • Apply SSF to the reference block; • Generate a prediction block for the current block; • Apply SSF to the prediction block.

[0166] (6) Intra-frame prediction (EIP) based on extrapolation filters EIP is an intra-frame prediction mode in which each pixel in the current block is predicted step by step based on the reconstructed neighboring pixels or the predicted neighboring pixels.

[0167] The formula for predicting blocks in EIP is as follows:

[0168]

[0169]

[0170] in, yes The set of all pixels within a rectangle, where the coordinates of the bottom right corner of the rectangle are... ,but Except for itself (e.g.) Figure 13 (As shown).

[0171] More details about EIP can be found in L. Xu, Y. Yu, H. Yu, D. Wang, EE2-1.14: An 基于外推滤波器的帧内预测模式 The document JVET-AG0058, JointVideo Experts Team (JVET), Jan. 2024, can be found there.

[0172] In intra-frame prediction based on extrapolation filters, SSF can be applied to the reconstructed values ​​of the template region, and the filtered samples are used to derive the relevant linear model:

[0173] The following formula is used to generate prediction blocks:

[0174]

[0175] SSF can also be used to predict blocks to generate a new prediction block:

[0176] In one embodiment for implementing a sample-side filter in EIP mode, the process may include: • Check if SSF is available and decode whether SSF is enabled for the current block; • Apply SSF to the reconstructed samples in the template region of the current block; • Calculate the prediction function coefficients for intra-frame prediction based on extrapolation filters; • Generate a prediction block for the current block; • Apply SSF to the prediction block.

[0177] Figure 14 A conceptual schematic of a device 600 is shown, which can be used to implement some embodiments of the present invention. Device 600 can be an encoder or a decoder. Device 600 can be a computer (e.g., a desktop computer, personal computer, tablet computer, etc.), a telephone, a PDA, or any other type of electronic device. The device includes various types of computer-readable storage media and interfaces for various other types of computer-readable storage media. Device 600 includes a processor 601 and a memory 602. Memory 602 is configured to store executable instructions that, when executed by the processor, cause the processor to perform any of the foregoing methods, including the intra-frame prediction method and the color component prediction method described above.

[0178] The processor 601 may be a single-core processor or a multi-core processor in different embodiments. In some embodiments, the processor may include a GPU, NPU, or DSP, which can offload various computing tasks or perform image processing procedures provided by the processor 601.

[0179] Some embodiments of this application also include electronic components, such as microprocessors, memory, and RAM, storing computer program instructions in a machine-readable or computer-readable medium (or, as referred to, a computer-readable storage medium, machine-readable medium, or machine-readable storage medium). Exemplary examples include RAM, ROM, read-only optical disc (CD-ROM), recordable optical disc (CD-R), rewritable optical disc (CD-RW), read-only digital versatile optical disc (e.g., DVD-ROM, dual-layer DVD-ROM), various recordable or rewritable DVDs (e.g., DVD-RAM, DVD-RW, DVD+RW, etc.), flash memory (e.g., SD card, mini-SD card, micro-SD card, etc.), magnetic and / or solid-state drives, read-only and recordable Blu-ray® optical discs, ultra-density optical discs, any other optical or magnetic media, and floppy disks. The computer-readable medium may store a computer program executable by at least one processing unit and includes a set of instructions for performing various operations. Examples of computer programs or computer code include machine code (e.g., code generated by a compiler) and files containing higher-level code executed by a computer, electronic component, or microprocessor using an interpreter.

[0180] While the above discussion primarily concerns microprocessors or multi-core processors that execute software, some embodiments can be executed by one or more integrated circuits, such as application-specific integrated circuits (ASICs) or field-programmable gate arrays (FPGAs). In some embodiments, the integrated circuit executes instructions stored on the circuit. Furthermore, some embodiments execute software stored in programmable logic devices (PLDs), ROM, or RAM devices.

[0181] As used in this specification and any claims, "computer," "server," "processor," and "memory" refer to electronic or other technical devices. These terms do not include persons or groups. As used in this specification and any claims, "computer-readable medium," "computer-readable media," and "machine-readable medium" are entirely limited to tangible, physical objects that store information in a computer-readable form. These terms do not include any wireless signals, wired download signals, or any other transient signals.

[0182] This application also provides a computer-readable medium configured to store executable instructions. When these instructions are executed by a processor, the processor can perform any of the foregoing methods and processes. Some of the aforementioned features and applications are implemented as software processes, which are specified as a set of instructions recorded on a computer-readable storage medium (also referred to as a computer-readable medium). When these instructions are executed by one or more computing or processing units (e.g., one or more processors, processor cores, or other processing units), the processing unit performs the operations indicated in the instructions. Exemplarily, computer-readable media include, but are not limited to, CD-ROMs, flash drives, random access memory (RAM) chips, hard disk drives, erasable programmable read-only memories (EPROMs), electrically erasable programmable read-only memories (EEPROMs), etc. Computer-readable media do not include carrier waves and electronic signals transmitted via wireless or wired connections.

[0183] In this application, "software" includes firmware stored in read-only memory or an application stored in magnetic storage that can be read into memory for processor processing. Furthermore, in some embodiments, multiple software inventions may be implemented as sub-parts of a larger program while still being distinct software inventions. In some embodiments, multiple software inventions may also be implemented as separate programs. Moreover, any combination of separate programs that collectively implement the software inventions described in this application is within the scope of protection claimed in this application. In some embodiments, when a software program is installed on one or more electronic systems and runs, one or more specific machine implementations are defined for executing and performing the operations of the software program.

[0184] This application has been described with numerous specific details, but those skilled in the art will understand that this application can be implemented in other specific forms without departing from its spirit. Furthermore, several accompanying drawings illustrate relevant processes and methods schematically. The specific operations within these processes need not be strictly performed in the order shown and described. The specific operations also need not be performed as a continuous sequence of operations; different specific operations may be performed in different embodiments. Further, the process can be implemented through several sub-processes or as part of a larger macro-process.

[0185] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Any equivalent structural or procedural modifications made based on the specification and drawings of this application, or any direct or indirect application thereto in other related technical fields, shall fall within the scope of protection of this application.

Claims

1. An intra-frame prediction method, characterized in that, The method includes: Obtain color component samples from the template region; The first sample-side filter (SSF) is applied to the color component samples; A linear function is obtained based on the filtered color component samples of the template region; the linear function is used to predict the chromaticity component based on the luminance component. Obtain the brightness sample of the current block; and The chromaticity prediction of the current block is generated based on the linear function and the luminance sample of the current block.

2. The method according to claim 1, characterized in that, After obtaining the brightness sample of the current block, the method further includes: Apply the second SSF to the brightness sample of the current block; The step of obtaining the chromaticity prediction of the current block based on the linear function and the luminance sample of the current block includes: generating the chromaticity prediction of the current block based on the linear function and the filtered luminance sample of the current block.

3. The method according to claim 2, characterized in that, The step of using the first SSF for the color component sample includes: The first SSF is applied to the brightness sample from the template region; The first SSF is the same as the second SSF.

4. The method according to claim 1, characterized in that, The method further includes: Apply the third SSF to the chromaticity prediction of the current block.

5. The method according to claim 1, characterized in that, The application of the first SSF to the color component sample includes at least one of the following: Apply the first SSF to the brightness sample of the template region; or The first SSF is applied to the chromaticity sample of the template region.

6. The method according to claim 1, characterized in that: The first SSF is a low-pass filter and a finite impulse response (FIR) filter.

7. The method according to claim 1, characterized in that, The method further includes: Determine if the first SSF is available; The step of applying the first SSF to the color component sample includes: applying the first SSF to the color component sample in response to an indication that the first SSF is available; Whether the first SSF is available is determined based on at least one of the following: One or more predefined prediction patterns; The size of the blocks in the template area; An indication of whether the first SSF from the encoder is available.

8. The method according to claim 1, characterized in that, The method further includes: Determine whether the first SSF is enabled; Whether the first sample-side filter is enabled is determined based on the following: an indication from the encoder indicating whether the first sample-side filter is enabled.

9. An intra-frame prediction method, characterized in that, The method includes: Obtain color component samples from the template region; A linear function is obtained based on the color component samples; the linear function is used to predict the chromaticity component based on the luminance component. Get the brightness sample of the current block; Use the second SSF for the brightness sample of the current block; and The chromaticity prediction of the current block is obtained based on the linear function and the filtered luminance sample of the current block.

10. The method according to claim 9, characterized in that, After obtaining color component samples from the template region, the method further includes: Use the first SSF for the color component sample; The step of obtaining a linear function based on the color component samples includes obtaining the linear function based on the filtered color component samples of the template region.

11. The method according to claim 9, characterized in that, The method further includes: The third SSF is used for chromaticity prediction of the current block.

12. The method according to claim 9, characterized in that: The second SSF is a low-pass filter and an FIR filter.

13. An intra-frame prediction method, characterized in that, The method includes: Obtain color component samples from the template region; Apply the first SSF to the color component sample; A linear function is obtained based on the filtered color component samples of the template region; the linear function is used to predict the chromaticity component based on the luminance component. Get the brightness sample of the current block; Apply the second sample-side filter (SSF) to the brightness sample of the current block; and Based on the linear function and the filtered luminance sample of the current block, the chromaticity prediction of the current block is obtained; Apply the third SSF to the chromaticity prediction of the current block.

14. The method according to claim 13, characterized in that: The first SSF, the second SSF, and the third SSF are low-pass filters and FIR filters, respectively.

15. The method according to claim 13, characterized in that, Using the first SSF for the color component sample includes at least one of the following: Apply the first SSF to the brightness sample of the template region; or The first SSF is applied to the chromaticity sample of the template region.

16. The method according to claim 15, characterized in that: The first SSF is applied to the brightness sample of the template region; and the first SSF and the second SSF are the same.

17. The method according to claim 13, characterized in that, The method further includes: Determine whether the first SSF, the second SSF, and the third SSF are available; Whether the first SSF, the second SSF, and the third SSF are available can be determined based on at least one of the following: One or more predefined prediction patterns; The size of the template area block; An indication of whether the first SSF from the encoder is available.

18. The method according to claim 13, characterized in that, The method further includes: Accept an instruction to enable at least one of the first SSF, the second SSF, or the third SSF.

19. A method for intra-frame or inter-frame prediction, characterized in that, The method includes: Obtain the sample-side filter (SSF); Identify one or more specific samples, wherein the specific samples are from at least one template region or the current block; Apply the SSF to the one or more specific samples; and Color component prediction is performed based on the filtered specific sample.

20. The method according to claim 19, characterized in that: The convolutional cross-component model CCCM is applied to the color component prediction. Applying the sample-side filter to the one or more specific samples includes at least one of the following: The sample-side filter is applied to the brightness sample in the template region; The sample-side filter is applied to the chroma sample in the template region; Apply the sample-side filter to the reconstructed brightness sample in the current block; or The sample-side filter is applied to the chromaticity prediction result of the color component prediction.

21. The method according to claim 19, characterized in that: The multi-model convolutional cross-component model (MM-CCCM) is applied to the color component prediction. Applying the sample-side filter to the one or more specific samples includes at least one of the following: The sample-side filter is applied to the luminance and / or chrominance samples in the template region, and the filtered specific samples are divided into two categories based on the average luminance value in the template region; or The pixels in the template region are divided into two categories based on the average brightness value in the template region, and the sample-side filter is used for the classified brightness samples and / or chromaticity samples in the template region.

22. The method according to claim 19, characterized in that: Apply the gradient linear model GLM or the cross-component linear model CCLM to the color component prediction; Applying the sample-side filter to the one or more specific samples includes: The sample-side filter is applied to the luminance and / or chrominance samples in the template region; Apply the sample-side filter to the reconstructed brightness sample in the current block; and The sample-side filter is applied to the chromaticity prediction result of the color component prediction.

23. The method according to claim 19, characterized in that: Local illumination compensation (LIC) is applied to the color component prediction; Applying the sample-side filter to the one or more specific samples includes: The sample-side filter is applied to the reconstructed samples in the template region of the reference block and / or the current block; The sample-side filter is applied to the predicted samples of the template region of the reference block; and The sample-side filter is applied to the predicted samples of the current block.

24. The method according to claim 19, characterized in that: The intra-frame prediction (EIP) based on extrapolation filters is applied to the color component prediction; The step of using the sample-side filter for the one or more specific samples includes: applying the sample-side filter to the reconstructed samples of the template region of the current block.

25. The method according to claim 19, characterized in that: The sample-side filters are low-pass filters and FIR filters.

26. A video encoding / decoding device, characterized in that, It includes a processor and a memory, the memory storing executable instructions that, when executed by the processor, implement the method as described in any one of claims 1 to 25.

27. A computer-readable medium, characterized in that, The system stores executable instructions that, when executed by a processor, implement the method as described in any one of claims 1 to 25.

Images