KR102585236B1 - Video encoding and decoding methods and devices - Google Patents

Abstract

This disclosure provides an encoding and decoding method and apparatus. The decoding method includes obtaining a predicted sample matrix of a current prediction block, and obtaining a sample matrix reconstructed according to the predicted sample matrix, and the values of elements of the predicted sample matrix are of the current image. It is obtained according to the values of samples in the filtered reconstructed sample matrix.

Description

Video encoding and decoding methods and devices

The present invention relates to the field of video encoding and decoding, and specifically, to intra block copy (IBC) in intra prediction mode.

In video encoding and decoding methods, an image is typically divided into a plurality of image blocks, and each image block is then encoded or decoded. For each image block, prediction, transformation, quantization, and entropy encoding are performed. The prediction step uses the values of reconstructed pixels of previously encoded image blocks (these pixels are referred to as reference pixels) to obtain predicted values. The current image block is predicted, and then the difference values between the actual and predicted values of the current image block are encoded into a bitstream. At the time of decoding, the decoder must also predict the current image block to be decoded to obtain prediction values using the values of the reconstructed pixels of previously decoded image blocks, and then decoded from the bitstream. The difference values are added to the prediction values to obtain reconstructed values of the decoded image block. To ensure consistency between encoding and decoding, the codec must use the same reference pixels and the same prediction method when performing prediction. There are many specific prediction methods, and the encoder selects a prediction method according to the current image block and then adds information about the selected prediction method to the bitstream so that the decoder can predict the current block using the same prediction method. The selected prediction method is notified to the decoder.

There is a need for an improved intra block copy method in intra prediction mode for use in video encoding and decoding.

According to an example embodiment of the present disclosure, obtaining a predicted sample matrix of a current prediction block; and obtaining a reconstructed sample matrix according to the predicted sample matrix, wherein the values of the elements of the predicted sample matrix are the values of the samples in the filtered reconstructed sample matrix of the current image. A decoding method obtained according to is provided.

This disclosure provides an improved intra block copy method in intra prediction mode for use in video encoding and decoding.

Figure 1A shows the steps of dividing an image into maximum coding units and a schematic diagram of the coding units.
1B shows QT (Quad Tree) partitioning, BT (Binary Tree) partitioning, and EQT (Extend Quad Tree) partitioning.
Figure 2 shows a schematic diagram of intra block copy mode.
Figure 3 shows a flowchart of a decoding method according to an embodiment of the present invention.
4 shows the filtered reconstructed sample matrix deblocking filtering for decoded predictively-compensated samples within the current largest coding unit, and prediction-compensated samples of the decoded largest coding unit. It shows a schematic diagram in which reference blocks are at different positions when including samples obtained by performing filtering and sample adaptive offset.
5 shows the filtered reconstructed sample matrix obtained by performing deblocking filtering on the decoded prediction-compensated samples in the current largest coding unit, and the prediction-compensated samples of the decoded largest coding unit. shows a schematic diagram in which reference blocks are at different positions when including samples obtained by performing deblocking filtering and sample adaptive offset on .
6 shows the filtered reconstructed sample matrix obtained by performing sample adaptive offset on the decoded prediction-compensated samples within the current largest coding unit, and the prediction-compensated samples of the decoded largest coding unit. When included, it shows a schematic diagram in which the reference blocks are at different positions.
Figure 7 shows a block diagram of a decoding device according to an embodiment of the present invention.
Figure 8 shows a flowchart of an encoding method according to an embodiment of the present invention.
Figure 9 shows a schematic diagram of samples of the boundary of a filtering block.
Figure 10 shows a block diagram of an encoding device according to an embodiment of the present invention.
Figure 11 shows a flowchart of a decoding method according to an embodiment of the present invention.
Figure 12 shows a block diagram of a decoding device according to an embodiment of the present invention.

According to an example embodiment of the present disclosure, obtaining a predicted sample matrix of a current prediction block; and obtaining a reconstructed sample matrix according to the predicted sample matrix, wherein values of elements of the predicted sample matrix are obtained according to values of samples in the filtered reconstructed sample matrix of the current image. This is provided.

Memory in which computer programs are stored; and a processor that performs the above-described decoding method when executing the computer programs.

determining an identifier indicating whether to use deblocking filtering for the current frame to be encoded depending on whether the current frame is screen content; and adding the identifier into a bitstream.

Memory in which computer programs are stored; and a processor that performs the above-described encoding method when executing the computer programs.

determining whether to perform deblocking filtering on prediction-compensated samples of the current coding unit depending on whether the current coding unit is screen content; And according to the result of the decision, performing deblocking filtering on the prediction-compensated samples of the current coding unit to obtain deblocking-filtered reconstructed samples. , a decoding method is provided.

Memory in which computer programs are stored; and a processor that performs the above-described decoding method when executing the computer programs.

A computer readable storage medium is provided that stores non-transitory computer readable instructions that, when executed by a computer, cause the computer to perform the methods described above. .

Hereinafter, various examples of the present disclosure will be described in more detail with reference to the accompanying drawings. However, the present disclosure may take different forms and should not be construed as limited to the embodiments described herein. Parts unrelated to this disclosure may be omitted for clarity. It should be noted that throughout the drawings, the same reference numbers are used to depict the same or similar elements, features, and structures.

The terms and vocabulary used in the following description are not limited to their literal meanings, but are used only to ensure that the present invention can be clearly and consistently understood. Accordingly, those skilled in the art to which the present invention pertains will understand that the following description of exemplary embodiments of the present invention is provided for illustrative purposes only and that the present invention, as defined by the claims and their equivalents, You must understand that this is not intended to be limiting.

It will be understood that the singular forms “a” and “an” include plural references, unless the context clearly dictates otherwise.

Figure 1A shows the steps of dividing an image into maximum coding units and a schematic diagram of the coding units. As shown in the left diagram of FIG. 1A, the image 100 is divided into a series of maximum coding units, and each maximum coding unit is decoded in turn according to raster scan order. One largest coding unit can be a single coding unit or be split into four smaller coding units with a quadtree partition structure. As shown in the right figure of FIG. 1A, the maximum coding unit 110 is divided into QT (Quad Tree) partitioning, BT (Binary Tree) partitioning, and EQT (Extend Quad Tree) partitioning. It can be divided into smaller coding units according to a basic block division structure, such as Tree)) division.

Figure 1B shows QT (quad tree) partitioning, BT (binary tree) partitioning, and EQT (extended quad tree) partitioning. Referring to FIG. 1B, QT divides coding unit 115 into four coding units. BT may split coding unit 120 into two left and right coding units, or BT may split coding unit 130 into two upper and lower coding units. EQT includes two (horizontal and vertical) I-shaped partition modes and splits the coding unit into four coding units. For example, as shown in FIG. 1B, EQT_VER divides coding unit 140 into four coding units, and EQT_HOR divides coding unit 150 into four coding units. A prediction unit includes a luminance prediction block and a corresponding chrominance prediction block, divided from the coding unit. If the coding unit type of the coding unit is 'I_2M_2N', 'I_2M_hN', 'I_2M_nU', 'I_2M_nD', 'I_hM_2N', 'I_nL_2N', or 'I_nR_2N', the prediction type of the coding unit is general intra prediction ( normal intra prediction), and the prediction type of the prediction blocks included in the coding unit is normal intra prediction. In contrast, when the coding unit type is 'IBC_2M_2N', the prediction type of the coding unit is intra block copy, and the prediction type of the prediction blocks included in the coding unit is intra block copy. In contrast, when the prediction type of the coding unit is inter prediction, the prediction type of the prediction units and prediction blocks included in the coding unit is inter prediction.

For screen content sequences, such as text and graphics, there are many repeated textures within the same frame, ie, there is a strong spatial correlation. If previously encoded regions of the current frame can be referenced when encoding the current block, coding efficiency can be greatly improved (targeting the strong spatial correlation characteristic of screen images). This technique, which refers to the encoded blocks of the current frame and is used for the prediction of the current block, is referred to as intra block copy (IBC).

Figure 2 shows a schematic diagram of intra block copy mode. Intra block copy is similar to inter image prediction, except that the prediction blocks of the intra block copy are generated using reconstructed blocks of the currently encoded image frame. For example, in intra block copy (IBC), the current block 210 is predicted using a reference block 220, which was previously processed prior to the current block 210 and is located within the same frame.

Deblocking filtering (DBF) is used to reduce the blocking effect. Sample adaptive offset (SAO) is used to improve the ringing effect. These two filters can effectively improve the subjective and objective quality of video.

In one embodiment, the values of the elements of the IBC-predicted sample matrix are obtained according to the values of the samples in the unfiltered reconstructed sample matrix of the current image. Deblocking filtering and sample adaptive offset processing may not be performed on the unfiltered reconstructed samples, which will result in large distortion of the values of the predicted samples and thus the quality and coding of the prediction blocks when using IBC. Efficiency decreases. In other words, when unfiltered reconstructed samples are used as reference samples for a prediction process using intra block copy (IBC), prediction/coding efficiency may deteriorate.

To improve prediction efficiency when performing prediction according to intra block copy (IBC), the present disclosure determines that the values of the elements of the IBC-predicted sample matrix depend on the values of the samples in the filtered reconstructed sample matrix of the current image. An intra block copy method is provided, which further improves the image coding efficiency of screen content.

Figure 3 shows a flowchart of a decoding method according to an embodiment of the present disclosure. The method includes the following steps: obtaining a predicted sample matrix of the current prediction block (S301); and obtaining a reconstructed sample matrix according to the predicted sample matrix, wherein values of elements of the predicted sample matrix are obtained according to values of samples in the filtered reconstructed sample matrix of the current image (S302). In this way, by using the filtered reconstructed sample matrix of the current image as a reference sample for IBC, the distortion of the values of predicted samples can be reduced, the quality of prediction blocks using IBC can be improved, and coding Efficiency can be improved.

The values of the elements of the predicted sample matrix may be obtained according to the values of the samples in the filtered reconstructed sample matrix of the current image. The values of the elements of the luminance predicted sample matrix may be obtained using the values of the samples in the filtered reconstructed sample matrix at full pixel precision of the current image, and the chrominance predicted sample matrix The values of the elements of the chrominance predicted sample matrix may be obtained using the values of the samples in the 1/2 accuracy filtered reconstructed chrominance sample matrix of the current image.

According to one embodiment, the filtered reconstructed sample matrix performs deblocking filtering and sample adaptive processing on decoded prediction-compensated samples within the current largest coding unit and prediction-compensated samples of the decoded largest coding unit. It may include samples obtained by performing offset. The prediction-compensated samples may be obtained using the sum of predicted samples and residual samples. In this way, distortion of the values of predicted samples can be reduced, the quality of prediction blocks using IBC can be improved, and coding efficiency can be improved.

4 shows the filtered reconstructed sample matrix performing deblocking filtering and sample adaptive offset on the decoded prediction-compensated samples within the current largest coding unit, and the prediction-compensated samples of the decoded largest coding unit. shows a schematic diagram in which the reference blocks are at different positions when including the samples obtained. In Figure 4, 5x5 maximum coding units are shown as an example.

For example, the filtered reconstructed sample matrix performs deblocking filtering and sample adaptive offset on the decoded prediction-compensated samples within the current largest coding unit, and the prediction-compensated samples of the decoded largest coding unit. When including samples obtained as follows, at least one of the following situations may exist. As shown in the left diagram 410 of FIG. 4, when the reference block 441 indicated by the block vector of the current block 431 is located completely outside the current largest coding unit (i.e., the reference block 441 If the current block 431 is located outside the current largest coding unit, and the reference block 441 is located inside the previously decoded largest coding unit), the values of the elements of the predicted sample matrix are all decoded. It is obtained according to the values of the samples obtained by performing deblocking filtering and sample adaptive offset on the prediction-compensated samples of the largest coding unit. As shown in the central diagram 420 of FIG. 4, a portion of the reference block 442 pointed to by the block vector of the current block 432 is outside the current largest coding unit (i.e., inside the decoded largest coding unit). ) and the remaining part of the reference block 442 is located inside the current maximum coding unit, some of the values of the elements of the predicted sample matrix are decoded with respect to the prediction-compensated samples of the decoded maximum coding unit. obtained according to the values of the samples obtained by performing blocking filtering and sample adaptive offset, and the remaining part of the values of the elements of the predicted sample matrix are obtained according to the values of the decoded prediction-compensated samples within the current largest coding unit. It is acquired. As shown in the right diagram 430 of FIG. 4, when the reference block 443 indicated by the block vector of the current block 433 is located completely inside the current maximum coding unit, the elements of the predicted sample matrix The values are all obtained according to the values of the decoded prediction-compensated samples within the current largest coding unit.

According to another embodiment, the filtered reconstructed sample matrix includes samples obtained by performing deblocking filtering on decoded prediction-compensated samples within the current largest coding unit, and prediction-compensated samples of the decoded largest coding unit. It may include samples obtained by performing deblocking filtering and sample adaptive offset on the obtained samples. In this way, distortion of the values of predicted samples can be reduced, the quality of prediction blocks using IBC can be improved, and coding efficiency can be improved.

5 shows the filtered reconstructed sample matrix obtained by performing deblocking filtering on the decoded prediction-compensated samples in the current largest coding unit, and the prediction-compensated samples of the decoded largest coding unit. shows a schematic diagram in which reference blocks are at different positions when including samples obtained by performing deblocking filtering and sample adaptive offset on .

For example, the filtered reconstructed sample matrix may include samples obtained by performing deblocking filtering on decoded prediction-compensated samples within the current largest coding unit, and prediction-compensated samples of the decoded largest coding unit. When including samples obtained by performing deblocking filtering and sample adaptive offset on , at least one of the following situations exists. As shown in the left diagram 510 of FIG. 5, the reference block 541 pointed to by the block vector of the current block 531 is completely outside the current largest coding unit (i.e., inside the decoded largest coding unit). ), the values of the elements of the predicted sample matrix all depend on the values of the samples obtained by performing deblocking filtering and sample adaptive offset on the prediction-compensated samples of the decoded largest coding unit. It is acquired. As shown in the central diagram 520 of FIG. 5, a portion of the reference block 542 pointed to by the block vector of the current block 532 is outside the current largest coding unit (i.e., the decoded largest coding unit). inside) and the remaining part of the reference block 542 is located inside the current largest coding unit, some of the values of the elements of the predicted sample matrix are in the prediction-compensated samples of the decoded largest coding unit. obtained according to the values of the samples obtained by performing deblocking filtering and sample adaptive offset on the decoded prediction-compensated samples in the current maximum coding unit. It is obtained according to the values of the samples obtained by performing deblocking filtering on . As shown in the right diagram 533 of FIG. 5, when the reference block 543 pointed by the block vector of the current block 533 is located completely inside the current maximum coding unit, the elements of the predicted sample matrix The values of are all obtained according to the values of the samples obtained by performing deblocking filtering on the decoded prediction-compensated samples in the current largest coding unit.

According to one embodiment, the values of the reconstructed samples may be deblocking-filtered or not deblocking-filtered, whether using deblocking filtering at the frame level, obtained from the code stream. It is indicated by the identifier. Therefore, for example, the decoding method further includes acquiring from the code stream an identifier indicating whether to use deblocking filtering for the current frame. For example, 0 indicates that no blocks in the current frame to be decoded use deblocking filtering, and 1 indicates that all blocks in the current frame to be decoded use deblocking filtering.

According to one embodiment, if the identifier indicates not to use deblocking filtering for the current frame (i.e., the identifier is 0), the filtered reconstructed sample matrix is within the current maximum coding unit. It includes decoded prediction-compensated samples, and samples obtained by performing sample adaptive offset on the prediction-compensated samples of the decoded largest coding unit. In this way, distortion of the values of predicted samples can be reduced, the quality of prediction blocks using IBC can be improved, and coding efficiency can be improved.

6 shows the filtered reconstructed sample matrix obtained by performing sample adaptive offset on the decoded prediction-compensated samples within the current largest coding unit, and the prediction-compensated samples of the decoded largest coding unit. When included, it shows a schematic diagram in which the reference blocks are at different positions.

For example, the filtered reconstructed sample matrix performs a sample adaptive offset on the decoded prediction-compensated samples within the current largest coding unit and the prediction-compensated samples of the decoded largest coding unit. When including acquired samples, at least one of the following situations exists. As shown in the left diagram 610 of FIG. 6, the reference block 651 pointed to by the block vector of the current block 641 is completely outside the current largest coding unit (i.e., inside the decoded largest coding unit). ), the values of the elements of the predicted sample matrix are all obtained according to the values of the samples obtained by performing sample adaptive offset on the prediction-compensated samples of the decoded largest coding unit. As shown in the central diagram 620 of FIG. 6, a portion of the reference block 652 pointed to by the block vector of the current block 642 is outside the current largest coding unit (i.e., the decoded largest coding unit). inside), and if the remaining part of the reference block 652 is located inside the current largest coding unit, some of the values of the elements of the predicted sample matrix are the prediction-compensated samples of the decoded largest coding unit. is obtained according to the values of the samples obtained by performing a sample adaptive offset on do. As shown in the right diagram 630 of FIG. 6, when the reference block 653 pointed by the block vector of the current block 643 is located completely inside the current largest coding unit, the elements of the predicted sample matrix The values of are all obtained according to the values of the decoded prediction-compensated samples within the current largest coding unit.

Information indicating whether deblocking filtering is applied to the current frame may be used. For example, an identifier with 0 or 1 can be used to indicate deblocking filtering. When the identifier indicates to use deblocking filtering for the current frame (eg, when the identifier is 1), two situations exist.

According to one embodiment, when the identifier indicates to use deblocking filtering for the current frame (e.g., when the identifier is 1), the filtered reconstructed sample matrix is used for decoding within the current largest coding unit. prediction-compensated samples, and samples obtained by performing deblocking filtering and sample adaptive offset on the prediction-compensated samples of the decoded largest coding unit. As described above, in this way, distortion of the values of predicted samples can be reduced, the quality of prediction blocks using IBC can be improved, and coding efficiency can be improved. Deblocking filtering is used at the frame level, but coding units within the current largest coding unit may not use deblocking filtering.

As described above, for example, the filtered reconstructed sample matrix deblocks for decoded prediction-compensated samples within the current largest coding unit and prediction-compensated samples of the decoded largest coding unit. When including samples obtained by performing filtering and sample adaptive offset, at least one of the following situations exists. As shown in the left diagram 410 of FIG. 4, the reference block 441 pointed to by the block vector of the current block 431 is completely outside the current largest coding unit (i.e., inside the decoded largest coding unit). ), the values of the elements of the predicted sample matrix all depend on the values of the samples obtained by performing deblocking filtering and sample adaptive offset on the prediction-compensated samples of the decoded maximum coding unit. It is acquired. As shown in the central diagram 420 of FIG. 4, a portion of the reference block 442 pointed to by the block vector of the current block 432 lies outside the current maximum coding unit (i.e., within the decoded maximum coding unit). unit) and the remaining part of the reference block 442 is located inside the current maximum coding unit, some of the values of the elements of the predicted sample matrix are prediction-compensated samples of the decoded maximum coding unit. is obtained according to the values of the samples obtained by performing deblocking filtering and sample adaptive offset on Obtained according to the values. As shown in the right diagram 430 of FIG. 4, when the reference block 443 indicated by the block vector of the current block 433 is located completely inside the current maximum coding unit, the predicted sample matrix The values of the elements are all obtained according to the values of the decoded prediction-compensated samples within the current largest coding unit.

According to another embodiment, when the identifier indicates to use deblocking filtering for the current frame (e.g., when the identifier is 1), the filtered reconstructed sample matrix is the decoded sample matrix within the current largest coding unit. Includes samples obtained by performing deblocking filtering on prediction-compensated samples, and samples obtained by performing deblocking filtering and sample adaptive offset on prediction-compensated samples of the decoded largest coding unit. do. As described above, in this way, distortion of the values of predicted samples can be reduced, the quality of prediction blocks using IBC can be improved, and coding efficiency can be improved.

As described above, for example, the filtered reconstructed sample matrix is obtained by performing deblocking filtering on decoded prediction-compensated samples within the current maximum coding unit, and the decoded maximum coding unit. When including samples obtained by performing deblocking filtering and sample adaptive offset on prediction compensated samples of a unit, at least one of the following situations exists. As shown in the left diagram 510 of FIG. 5, the reference block 541 pointed to by the block vector of the current block 531 is completely outside the current largest coding unit (i.e., inside the decoded largest coding unit). ), the values of the elements of the predicted sample matrix all depend on the values of the samples obtained by performing deblocking filtering and sample adaptive offset on the prediction-compensated samples of the decoded maximum coding unit. It is acquired. As shown in the central diagram 520 of FIG. 5, a portion of the reference block 542 pointed to by the block vector of the current block 532 is outside the current largest coding unit (i.e., the decoded largest coding unit). inside) and the remaining part of the reference block 542 is located inside the current largest coding unit, some of the values of the elements of the predicted sample matrix are in the prediction-compensated samples of the decoded largest coding unit. obtained according to the values of the samples obtained by performing deblocking filtering and sample adaptive offset on It is obtained according to the values of the samples obtained by performing deblocking filtering. As shown in the right diagram 533 of FIG. 5, when the reference block 543 indicated by the block vector of the current block 533 is located completely inside the current maximum coding unit, the predicted sample matrix The values of the elements are all obtained according to the values of the samples obtained by performing deblocking filtering on the decoded prediction-compensated samples within the current largest coding unit.

Figure 7 shows a block diagram of a decoding device according to an embodiment of the present disclosure. The decoding device 700 includes a memory 701 in which computer programs are stored, and a processor 702 that performs the above-described decoding methods when executing the computer programs. Such a device can reduce distortion of predicted sample values, improve the quality of prediction blocks using IBC, and improve coding efficiency.

Figure 8 shows a flowchart of an encoding method according to an embodiment of the present disclosure. This embodiment relates to a method of obtaining an identifier of whether to use deblocking filtering at the frame level, as described above. The method includes determining an identifier indicating whether to use deblocking filtering for the current frame to be encoded depending on whether the current frame to be encoded is screen content (step S801); and writing the identifier into a code stream (step S802). In this way, coding efficiency can be improved.

The screen content may be captured directly from an image display unit of a device, such as a computer or mobile terminal, such as computer graphics, text images, images of a mixture of natural images and graphics and characters, and computer-generated animated images, etc. , it is an image. Screen content like this is common in application scenarios, such as desktop collaboration, desktop sharing, and cloud gaming. The main difference between screen content images and natural images captured by a camera is that screen content images are noise-free and have discrete tones and sharp edges, whereas natural images typically contain noise and have continuous images. It has continuous tones and complex textures. The calculation method for determining whether an image is screen content may be one or other of the following methods and is not limited to this specification. One possible implementation is to construct a hash table for each frame before encoding, and for each hash value, calculate whether the value corresponds to the presence of two or more blocks, and determine the probability of occurrence of the situation. When greater than this certain threshold, the current frame is considered screen content. Another possible implementation is to convolve each frame with a sobel operator before encoding, and when the calculated Sobel value of the entire frame is greater than a certain threshold, the current frame is considered screen content.

According to one embodiment, step S801 sets the identifier to indicate that deblocking filtering is not used for the current frame when the current frame to be encoded is screen content (e.g., the identifier is set to 0). set), or, if the current frame to be encoded is not screen content, the identifier is set to indicate that deblocking filtering is used for the current frame (e.g., the identifier is set to 1).

According to one embodiment, the encoding method further includes determining, based on the identifier, whether to perform deblocking filtering on prediction-compensated samples of the encoded current frame. When the identifier indicates that deblocking filtering is not used for the current frame (e.g., when the identifier is 0), deblocking filtering is not used for filtering blocks of the current frame, and the identifier When indicates to use deblocking filtering for the current frame (e.g., when the identifier is 1), it is determined whether to use deblocking filtering for the filtering blocks of the current frame.

According to one embodiment, based on the identifier, determining whether to perform deblocking filtering on prediction-compensated samples of the current frame to be encoded includes filter strength of the boundary of the filtering block. strength), and calculating the filter strength of the boundary of the filtering block includes at least one of the following steps. When it is determined that the boundary should be filtered, if the absolute value of (p0-q0) is greater than or equal to 4 times the first parameter, the filter strength is 0, where p0 is the first parameter to the left or above the boundary. sample, q0 is the first sample to the right or below the boundary, and the first parameter is calculated according to the quantization parameter and bitdepth, or when the smoothness of the boundary is 6 , the absolute value of (p0-p1) is less than or equal to 1/4 of the second parameter, the absolute value of (q0-q1) is less than or equal to 1/4 of the second parameter, and the absolute value of (p0-p3) is less than or equal to 1/4 of the second parameter. The value is less than or equal to 1/2 of the second parameter, the absolute value of (q0-q3) is less than or equal to 1/2 of the second parameter, and the absolute value of (p0-q0) is less than the first parameter. When small, the filter strength is 4, where p1 is the second sample to the left or above the boundary, q1 is the second sample to the right or below the boundary, and p3 is the fourth sample to the left or above the boundary. sample, q3 is the fourth sample to the right or below the boundary, and the second parameter is calculated according to the quantization parameter and the bit depth, or when the smoothness of the boundary is 5, p0 is equal to p1 and q0 is equal to q1 and the absolute value of (p2-q2) is less than the first parameter, then the filter strength is 3, where p2 is the third sample to the left or above the boundary and q2 is the third sample to the right or above the boundary. This is the third sample on the lower side.

Figure 9 shows a schematic diagram of samples of the boundary of a filtering block. In Figure 9, the black thick line represents a specific segment of the boundary of the filtering block. The eight samples on either side of the boundary of the filtering block are denoted p0, p1, p2, p3 and q0, q1, q2, q3, respectively, i.e., p0 is the first sample to the left or above the boundary, p1 is the second sample to the left or above the boundary, p2 is the third sample to the left or above the boundary, p3 is the fourth sample to the left or above the boundary, and q0 is the right or below the boundary. is the first sample, q1 is the second sample to the right or below the boundary, q2 is the third sample to the right or below the boundary, and q3 is the fourth sample to the right or below the boundary. For each boundary of the filtering block, it is determined whether the boundary should be filtered. When the boundary is to be filtered, the filter strength of each segment of the boundary is calculated, and deblocking filtering is performed according to the filter strength of the segment of the boundary. Otherwise, the value of the compensated sample is taken directly as the value of the deblocking-filtered sample.

For example, the derivation process of the boundary deblocking filter strength Bs is as follows. The larger the value of the boundary filter strength Bs, the greater the filter strength. Bs of 0 indicates no filtering.

When it is determined that filtering is necessary, if the absolute value of (p0-q0) is greater than or equal to 4 times the first parameter, Bs is 0, where the first parameter is a quantization parameter (QP) and a bit Calculated based on depth. Otherwise, the process of calculating the filter strength Bs of the boundary is as follows.

(1) Calculating the smoothness fS of the boundary:

If the absolute value of (p0-p1) is less than the second parameter, the left/top smoothness of the boundary is increased by 2, and the second parameter is calculated according to the quantization parameter and the bit depth. If the absolute value of (p0-p2) is smaller than the second parameter, the left/top smoothness of the boundary is further increased by 1. Right/bottom smoothness is obtained similarly. The smoothness fS of the boundary is equal to the sum of the left/upper side smoothness and the right/lower side smoothness.

(2) Obtaining Bs according to the value of the boundary smoothness fS:

When fS is 6, the absolute value of (p0-p1) is less than or equal to 1/4 of the second parameter, the absolute value of (q0-q1) is less than or equal to 1/4 of the second parameter, and ( The absolute value of (p0-p3) is less than or equal to 1/2 of the second parameter, the absolute value of (q0-q3) is less than or equal to 1/2 of the second parameter, and the absolute value of (p0-q0) If this is less than the first parameter, Bs is 4, otherwise, Bs is 3.

When fS is 5, if p0 is equal to p1, q0 is equal to q1, and the absolute value of (p2-q2) is less than the first parameter, Bs is 3, otherwise, Bs is 2.

Figure 10 shows a block diagram of an encoding device according to an embodiment of the present disclosure. The encoding device 1000 includes a memory 1001 in which computer programs are stored, and a processor 1002 that performs the above-described encoding method when executing the computer programs. The encoding device can improve coding efficiency.

Figure 11 shows a flowchart of a decoding method according to an embodiment of the present disclosure. The method concerns whether to use deblocking filtering at the block level. The method includes determining whether to perform deblocking filtering on prediction-compensated samples of the current coding unit depending on whether the current coding unit is screen content (S1101), and according to the result of the decision, deblocking filtering. and performing deblocking filtering on the prediction-compensated samples of the current coding unit to obtain blocking-filtered reconstructed samples (S1102). In this way, the quality of the decoded image can be improved.

According to one embodiment, step S1101 is, if the currently decoded block is not screen content, determining to use deblocking filtering on the values of the reconstructed samples of the currently decoded block, or When the decoded block is screen content, determining not to use deblocking filtering on values of reconstructed samples of the currently decoded block.

The current coding unit may be the largest decoding block or a decoding block, but is not limited to this specification.

The calculation method for determining whether the currently decoded block is screen content may be one or other of the following methods and is not limited to this specification. One possible implementation is to convolve the currently decoded block with a Sobel operator, and when the calculated Sobel value of the currently decoded block is greater than a certain threshold, the currently decoded block is screen content. Another possible implementation is to calculate a histogram of the currently decoded block, and when the calculated number of sample value types of the currently decoded block is less than a certain threshold, the currently decoded block is screen content.

Figure 12 shows a block diagram of a decoding device according to an embodiment of the present disclosure. The decoding device 1200 includes a memory 1201 in which computer programs are stored, and a processor 1202 that performs the decoding method described with reference to FIG. 11 when the computer programs are executed. The decoding device can improve the quality of decoded images.

The present disclosure also provides a computer readable storage medium that, when executed by a computer, stores non-transitory computer readable instructions that cause the computer to perform the methods described above. ) is provided.

The exemplary embodiments described herein are not intended to be limiting. Aspects of the disclosure, as generally described herein and shown in the drawings, can be arranged, replaced, combined, separated, and designed in a variety of different configurations. may be designed, all of which are contemplated herein. Additionally, the features shown in each drawing may be used in combination with each other, unless otherwise indicated by context. Accordingly, the drawings should generally be regarded as a constituent part of one or more overall embodiments, but it should be understood that not all of the features shown are essential to each embodiment.

Although the present disclosure has been illustrated and described with reference to specific implementations, various changes and modifications may be made by those skilled in the art to which the present disclosure pertains without departing from the spirit and scope of the present disclosure as defined by the claims. You will understand.

Claims (16)

delete delete delete delete delete delete delete delete Generating a bitstream including an identifier indicating whether to perform deblocking filtering on the current frame,
If the identifier indicates that deblocking filtering is to be performed:
When the absolute value of (p0-q0) is greater than or equal to 4 times the first parameter, the filter strength is determined to be 0, where p0 is the first sample to the left or above the boundary, and q0 is the right or upper side of the boundary. It is the first sample on the lower side, and the first parameter is calculated according to the quantization parameter and bitdepth,
Based on the absolute value of (p0-p3) being less than or equal to 1/2 of the second parameter and the absolute value of (q0-q3) being less than or equal to 1/2 of the second parameter, the filter strength is adjusted to 4. where p3 is the fourth sample to the left or above the boundary, q3 is the fourth sample to the right or below the boundary, and the second parameter is calculated according to the quantization parameter and the bit depth,
Based on the absolute value of (p2-q2) being less than the first parameter, the filter strength is determined to be 3, where p2 is the third sample to the left or above the boundary and q2 is the third sample to the right or below the boundary. Third sample, encoding method. delete delete The method of claim 9, wherein when the identifier indicates that deblocking filtering is performed:
When the absolute value of (p0-q0) is greater than or equal to 4 times the first parameter, determining the filter strength to be 0, where p0 is the first sample to the left or above the boundary, and q0 is the first sample to the left or above the boundary. a first sample on the right or bottom, wherein the first parameter is calculated according to a quantization parameter and a bitdepth;
When the smoothness of the boundary is 6, the absolute value of (p0-p1) is less than or equal to 1/4 of the second parameter and the absolute value of (q0-q1) is less than 1/4 of the second parameter. is less than or equal to, the absolute value of (p0-p3) is less than or equal to 1/2 of the second parameter, the absolute value of (q0-q3) is less than or equal to 1/2 of the second parameter, and the (p0 If the absolute value of -q0) is less than the first parameter, determining the filter strength to be 4, where p1 is the second sample to the left or above the boundary, and q1 is the second sample to the right or below the boundary. 2 samples, p3 is the fourth sample to the left or above the boundary, q3 is the fourth sample to the right or below the boundary, and the second parameter is calculated according to the quantization parameter and the bit depth. ; and
When the smoothness of the boundary is 5, p0 is equal to p1, q0 is equal to q1, and the absolute value of (p2-q2) is less than the first parameter, determining the filter strength to be 3, p2 is a third sample to the left or above the boundary and q2 is a third sample to the right or below the boundary.
memory in which computer programs are stored, and
A processor that performs an encoding method when executing the computer programs, the method comprising:
Generating a bitstream including an identifier indicating whether to perform deblocking filtering on the current frame,
If the identifier indicates that deblocking filtering is to be performed:
When the absolute value of (p0-q0) is greater than or equal to 4 times the first parameter, the filter strength is determined to be 0, where p0 is the first sample to the left or above the boundary, and q0 is the right or upper side of the boundary. The lower first sample, wherein the first parameter is calculated according to the quantization parameter and the bit depth,
Based on the absolute value of (p0-p3) being less than or equal to 1/2 of the second parameter and the absolute value of (q0-q3) being less than or equal to 1/2 of the second parameter, the filter strength is adjusted to 4. where p3 is the fourth sample to the left or above the boundary, q3 is the fourth sample to the right or below the boundary, and the second parameter is calculated according to the quantization parameter and the bit depth,
Based on the absolute value of (p2-q2) being less than the first parameter, the filter strength is determined to be 3, where p2 is the third sample to the left or above the boundary and q2 is the third sample to the right or below the boundary. Third sample, encoding device.
Obtaining an identifier indicating whether to perform deblocking filtering on prediction-compensated samples of the current coding unit,
If the identifier indicates that deblocking filtering is to be performed:
When the absolute value of (p0-q0) is greater than or equal to 4 times the first parameter, the filter strength is determined to be 0, where p0 is the first sample to the left or above the boundary, and q0 is the right or upper side of the boundary. The lower first sample, wherein the first parameter is calculated according to the quantization parameter and the bit depth,
Based on the absolute value of (p0-p3) being less than or equal to 1/2 of the second parameter and the absolute value of (q0-q3) being less than or equal to 1/2 of the second parameter, the filter strength is adjusted to 4. where p3 is the fourth sample to the left or above the boundary, q3 is the fourth sample to the right or below the boundary, and the second parameter is calculated according to the quantization parameter and the bit depth,
Based on the absolute value of (p2-q2) being less than the first parameter, the filter strength is determined to be 3, where p2 is the third sample to the left or above the boundary and q2 is the third sample to the right or below the boundary. Third sample, decoding method.
The method of claim 14, wherein when the identifier indicates that deblocking filtering is performed:
When the absolute value of (p0-q0) is greater than or equal to 4 times the first parameter, determining the filter strength to be 0, where p0 is the first sample to the left or above the boundary, and q0 is the first sample to the left or above the boundary. a first sample on the right or bottom, wherein the first parameter is calculated according to a quantization parameter and a bit depth;
When the smoothness of the boundary is 6, the absolute value of (p0-p1) is less than or equal to 1/4 of the second parameter and the absolute value of (q0-q1) is less than or equal to 1/4 of the second parameter. The absolute value of (p0-p3) is less than or equal to 1/2 of the second parameter, the absolute value of (q0-q3) is less than or equal to 1/2 of the second parameter, and the absolute value of (p0-q0) is less than or equal to 1/2 of the second parameter. If the absolute value of is less than the first parameter, determining the filter strength to be 4, where p1 is the second sample to the left or above the boundary, q1 is the second sample to the right or below the boundary, and , p3 is the fourth sample to the left or above the boundary, q3 is the fourth sample to the right or below the boundary, and the second parameter is calculated according to the quantization parameter and the bit depth; and
When the smoothness of the boundary is 5, p0 is equal to p1, q0 is equal to q1, and the absolute value of (p2-q2) is less than the first parameter, determining the filter strength to be 3, p2 is a third sample to the left or above the boundary and q2 is a third sample to the right or below the boundary.
memory in which computer programs are stored, and
A processor that performs an encoding method when executing the computer programs, the method comprising:
Obtaining an identifier indicating whether to perform deblocking filtering on prediction-compensated samples of the current coding unit,
If the identifier indicates that deblocking filtering is to be performed:
When the absolute value of (p0-q0) is greater than or equal to 4 times the first parameter, the filter strength is determined to be 0, where p0 is the first sample to the left or above the boundary, and q0 is the right or upper side of the boundary. The lower first sample, wherein the first parameter is calculated according to the quantization parameter and the bit depth,
Based on the absolute value of (p0-p3) being less than or equal to 1/2 of the second parameter and the absolute value of (q0-q3) being less than or equal to 1/2 of the second parameter, the filter strength is adjusted to 4. where p3 is the fourth sample to the left or above the boundary, q3 is the fourth sample to the right or below the boundary, and the second parameter is calculated according to the quantization parameter and the bit depth,
Based on the absolute value of (p2-q2) being less than the first parameter, the filter strength is determined to be 3, where p2 is the third sample to the left or above the boundary and q2 is the third sample to the right or below the boundary. A third sample, a decoding device.

Images