CN103338373A - Adjacent boundary length deducing method and device - Google Patents

Abstract

An embodiment of the present invention provides a method and device for deriving adjacent boundary lengths. The adjacent boundary length derivation method includes: determining whether to merge consecutive adjacent blocks into one adjacent block according to whether the prediction information in consecutive adjacent blocks is consistent. unit and derive the adjacent unit boundary information, and calculate the adjacent boundary length according to the boundary information of the adjacent unit and the current block adjacent edge. Through the method for calculating the adjacent boundary length, the adjacent boundary length information can be calculated while saving memory overhead. The obtained adjacent boundary length information can be applied to subsequent video coding and decoding technologies such as motion vector prediction and intra-frame most probable mode prediction to improve video coding efficiency.

Description

A method and device for deriving adjacent boundary length

technical field

The present invention relates to a video encoding and decoding technology, in particular to a method and a device for deriving adjacent boundary lengths in video encoding and decoding.

Background technique

In video coding and decoding technology, the length of the adjacent boundary reflects the strength of the correlation between two adjacent blocks. Using this correlation, we can predict the coding information of the current block through the coded adjacent blocks. This enables adjacent boundary lengths to play a role in video codecs. The motion vector prediction method based on the length of the adjacent boundary is to select the motion vector of the block with the longest adjacent boundary with the current pixel among the adjacent blocks of the current pixel block as the motion vector predictor of the current block. Compared with the technology of AVS MEDIAN in the original video coding standard AVS, the motion vector prediction method based on the adjacent boundary length can obtain better performance.

However, in the original motion vector prediction method based on the adjacent boundary length, the adjacent boundary is derived according to the boundary coordinate position of the PU (prediction unit) block adjacent to the current block and the coordinate position of the current block boundary. That is, the length of the overlapping boundary between the adjacent PU block and the current pixel block is the length of the adjacent boundary. This derivation method needs to store additional CU (coding unit) block size information and block division information of the CU block in advance, which increases memory overhead.

Contents of the invention

The purpose of the present invention is to address the deficiencies in the prior art, and proposes to decide whether to merge the continuous adjacent blocks into one adjacent unit according to whether the prediction information in the continuous adjacent blocks is consistent and derive the boundary information of the adjacent units, and then according to the boundary Information on how to calculate the length of adjacent boundaries. Through the method for calculating the adjacent boundary length, the adjacent boundary length information can be calculated while saving memory overhead. The obtained adjacent boundary length information can be applied to subsequent video coding and decoding technologies such as motion vector prediction and intra-frame most probable mode prediction to improve video coding efficiency.

The first purpose of the present invention is to provide a method for deriving adjacent boundary lengths, which includes the following steps:

Obtain at least two adjacent blocks of the current pixel block;

Merge adjacent blocks whose adjacent blocks are continuous and whose prediction information is consistent into one adjacent unit;

Calculate the length of the overlapping boundary between the adjacent unit and the current pixel block, and the resulting length is the length of the adjacent boundary between the adjacent unit and the current pixel block.

Further, the consistent prediction information in the continuous adjacent blocks means that the motion information of inter-frame prediction blocks is the same or the deviation is within a small range.

Further, the consistent prediction information in the consecutive adjacent blocks means that the intra prediction modes of the intra prediction blocks are the same or the direction deviation is within a small range.

Further, the motion information of the inter-prediction block is: the motion vector of the inter-prediction block or the motion vector of the inter-prediction block according to the The value obtained after distance scaling.

Another object of the present invention is to provide a device for deriving adjacent boundary lengths, which is characterized in that it includes:

An input module for obtaining at least two adjacent blocks of the current pixel block;

An adjacent unit fusion module for merging adjacent blocks whose adjacent blocks are continuous and whose prediction information is consistent into one adjacent unit;

The adjacent boundary length calculation module is used to calculate the length of the overlapping boundary between the adjacent unit and the current pixel block, and the resulting length is the length of the adjacent boundary between the adjacent unit and the current pixel block.

Further, the consistent prediction information in the continuous adjacent blocks means that the motion vectors of inter-frame prediction blocks are the same or the deviation is within a small range.

Further, the consistent prediction information in consecutive adjacent blocks means that the intra prediction modes of the intra prediction blocks are the same or the direction deviation is within a small range.

Further, the motion information of the inter-prediction block is: the motion vector of the inter-prediction block or the motion vector of the inter-prediction block according to the The value obtained after distance scaling.

The beneficial effect of the present invention is that the adjacent adjacent boundary length derivation method and device provided by the embodiment of the present invention judge whether the consecutive adjacent blocks belong to the same adjacent unit according to whether the prediction information in the consecutive adjacent blocks is the same, and derive the Adjacent unit boundary information is described, and the adjacent boundary length is calculated according to the boundary information. According to the obtained adjacent boundary length information, it can be applied to subsequent video coding and decoding technologies such as motion vector prediction and intra-frame most probable mode prediction to improve video coding efficiency. .

Description of drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings that need to be used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description These are some embodiments of the present invention. For those skilled in the art, other drawings can also be obtained according to these drawings without any creative effort.

FIG. 1 is a schematic diagram of a current inter-frame pixel block and its adjacent blocks provided by an embodiment of the present invention;

FIG. 2 is a schematic diagram of an inter-frame current pixel block and its adjacent motion vector storage unit provided by an embodiment of the present invention;

FIG. 3 is a schematic diagram of another inter-frame current pixel block and its adjacent blocks provided by an embodiment of the present invention;

FIG. 4 is a schematic diagram of a current pixel block in a frame and its surrounding adjacent blocks provided by an embodiment of the present invention;

FIG. 5 is a schematic structural diagram of an apparatus for deriving adjacent boundary information in video coding and decoding according to an embodiment of the present invention.

Detailed ways

In inter prediction, video data can be divided into pixel blocks of different sizes for motion estimation. For example, 4X4, 4X8, 8X4, 8X8, 4X16, 8X16, 12X16, 16X16, 16X32, 32X32, 32X64, 64X64, etc. In intra prediction, video data can be divided into 64X64, 32X32, 16X16, 8X8, etc. A pixel block to be encoded or decoded is called a pixel block to be processed; a pixel block that has been encoded or decoded is called a processed pixel block. There may be multiple coded pixel blocks around the pixel block to be coded, and among these coded pixel blocks, there are pixel blocks in intra-frame coding mode and pixel blocks in inter-frame coding mode.

In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments It is a part of embodiments of the present invention, but not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without creative efforts fall within the protection scope of the present invention.

Example 1

FIG. 1 is a schematic diagram of a current pixel block and its adjacent pixel blocks in the spatial domain.

This example provides a method for deriving the adjacent boundary length of inter-frame prediction blocks in video encoding and decoding, including:

Step 1, at least two adjacent blocks of the current pixel block are obtained.

Specifically, as shown in FIG. 1 , adjacent blocks in the spatial domain of the current pixel block include A1, A2, ... , AN, B1, B2, ... BM, where N and M are natural numbers greater than 0. Among them, K blocks are inter-frame prediction blocks, 0≤K≤N+M. The motion information of the block in the same direction as the reference frame of the current pixel block among the K inter-frame prediction blocks is stored in the input device.

Step 2: Merge adjacent blocks whose adjacent blocks are continuous and whose prediction information is consistent into one adjacent unit.

Specifically, a possible implementation method is to compare whether the motion vectors in consecutive adjacent blocks in A1, A2 ... AN, B1, B2, ... BM are the same, if the motion vectors of any consecutive adjacent blocks are the same , then merge them into one adjacent unit. A neighboring block counts itself as a neighboring unit if its motion vector is different from any neighboring blocks connected to it. According to the method, several adjacent units are obtained, denoted as P1, P2, P3....

Step 3: Calculate the length of the overlapping boundary between the adjacent unit and the current pixel block, and the obtained length is the length of the adjacent boundary between the adjacent unit and the current pixel block.

Specifically, a possible implementation method is to calculate the length of the coincident boundary between the adjacent unit and the current pixel block according to the coordinate positions of the boundary points P1, P2, P3... and the coordinate positions of the current pixel block boundary points, and the obtained The length is the adjacent boundary length between the adjacent unit and the current pixel block.

The subsequent processing is not included in the patent claims, but it is added to the embodiments in order to make the embodiments clearer and more complete.

Specifically, among the adjacent units, the adjacent unit with the longest border with the current pixel block is used as the reference adjacent unit. The motion vector in this adjacent unit is taken out as the motion vector predictor or motion vector or motion vector predictor candidate or motion vector candidate of the current pixel block. If the motion vector predictor or motion vector is fetched, the motion vector of the reference adjacent block is directly used as the motion vector predictor or motion vector of the current block. If the motion vector predictor candidate or motion vector candidate is fetched, put the motion vector of the reference adjacent block into the motion vector predictor candidate or motion vector candidate list of the current block. After obtaining the motion vector predictor candidate or the motion vector predictor or the motion vector predictor or the motion vector, corresponding post-processing needs to be performed on the motion vector predictor candidate or the motion vector candidate or the motion vector predictor or the motion vector. Retain the same motion vector as the reference frame of the current pixel block in the motion vector of the neighboring block in the space domain; for the motion vector in the motion vector of the adjacent block in the space domain, which is in the same direction as the reference frame of the current pixel block but different from the reference frame, do the following processing, The motion vector×the distance between the current frame and its reference frame÷the distance between the frame where the motion vector is located and the reference frame.

Example 2

FIG. 2 is a schematic diagram of a current pixel block and its adjacent motion vector storage units in the spatial domain.

This example provides a method for deriving adjacent boundaries of inter-frame prediction blocks in video encoding and decoding, including:

Step 1, at least two adjacent blocks of the current pixel block are obtained.

Specifically, as shown in FIG. 2, there are motion vector storage units A1-A4, B1-B4, and C1-C3 around the current pixel block. These motion vector storage units are said adjacent blocks. First obtain the motion vectors in these motion vector storage units.

Step 2: Merge adjacent blocks whose adjacent blocks are continuous and whose prediction information is consistent into one adjacent unit.

Specifically, after obtaining a plurality of adjacent blocks in step 1, check whether the blocks in A1-A4 are consecutive blocks with the same motion vector, and fuse the consecutive blocks with the same motion vector into one adjacent unit. Check whether the blocks in B1-B4 are continuous blocks with the same motion vector, and fuse the continuous blocks with the same motion vector into one adjacent unit. A neighboring block counts itself as a neighboring unit if its motion vector is different from any neighboring blocks connected to it. Neighboring units P1, P2, P3... are obtained.

Step 3: Calculate the length of the overlapping boundary between the adjacent unit and the current pixel block, and the obtained length is the length of the adjacent boundary between the adjacent unit and the current pixel block.

Specifically, a possible implementation method is to calculate the number of motion vector storage units adjacent to the current pixel block included in P1, P2, P3..., and the obtained number is the number of adjacent units to the current pixel block border length.

The subsequent processing is not included in the patent claims, but it is added to the embodiments in order to make the embodiments clearer and more complete.

Specifically, among the adjacent units, the two adjacent units with the longest boundaries adjacent to the current pixel block are used as reference adjacent units. The motion vectors in these two adjacent units are taken out as motion vector predictor candidates or motion vector candidates. Putting the motion vector of the reference neighboring unit into a motion vector predictor candidate or a motion vector candidate list. After the motion vector predictor candidate or the motion vector candidate is obtained, corresponding post-processing needs to be performed on the motion vector predictor candidate or the motion vector candidate. Keep the same motion vector as the reference frame of the current pixel block in the motion vector of the adjacent block in the space domain; for the motion vector in the motion vector of the reference adjacent block in the space domain, which is in the same direction as the reference frame of the current pixel block but different from the reference frame, do the following processing , the motion vector × the distance between the current frame and its reference frame ÷ the distance between the frame where the motion vector is located and the reference frame.

Example 3

FIG. 3 is a schematic diagram of a current pixel block and its neighboring blocks in the spatial domain.

This example provides a method for deriving adjacent boundaries of inter-frame prediction blocks in video encoding and decoding, including:

Step 1, at least two adjacent blocks of the current pixel block are obtained.

Specifically, as shown in FIG. 3 , there are six adjacent blocks A1, A2, A3, B1, B2, and B3 around the current pixel block. Motion vectors in these neighboring blocks are obtained first.

Step 2: Merge adjacent blocks whose adjacent blocks are continuous and whose prediction information is consistent into one adjacent unit.

Specifically, a possible implementation method is to compare whether the motion vectors in consecutive adjacent blocks in A1, A2, and A3 are the same, compare whether the motion vectors in consecutive adjacent blocks in B1, B2, and B3 are the same, and if any If the motion vectors of consecutive adjacent blocks are the same, they are merged into one adjacent unit. A neighboring block counts itself as a neighboring unit if its motion vector is different from any neighboring blocks connected to it. According to the method, several adjacent units are obtained, denoted as P1, P2, P3.... For example: a situation can be that the motion vectors in two consecutive adjacent blocks of A2 and A3 are the same, which is recorded as P1; the motion vectors in the two consecutive adjacent blocks of B1 and B2 are the same, which is recorded as P2, and A1 is recorded as P3 and B3 are recorded as P4. It may also be that the motion vectors of A1, A2, and A3 are the same, which is marked as the adjacent unit P1, and the motion vectors of B1, B2, and B3 are the same, which is marked as the adjacent unit P2.

Step 3: Calculate the length of the overlapping boundary between the adjacent unit and the current pixel block, and the obtained length is the length of the adjacent boundary between the adjacent unit and the current pixel block.

Specifically, a possible implementation method is to calculate the length of the coincident boundary between the adjacent unit and the current pixel block according to the coordinate positions of the boundary points P1, P2, P3... and the coordinate positions of the current pixel block boundary points, and the obtained The length is the adjacent boundary length between the adjacent unit and the current pixel block.

The subsequent processing is the same as that described in the last paragraph of Embodiment 1, and will not be repeated here.

Example 4

FIG. 4 is a schematic diagram of a current pixel block and its neighboring blocks in the spatial domain.

This example provides a method for deriving adjacent boundaries of intra prediction blocks in video encoding and decoding, including:

Step 1, at least two adjacent blocks of the current pixel block are obtained.

Specifically, as shown in FIG. 4, there are six adjacent blocks A1, A2, A3, B1, B2, and B3 around the current pixel block. Firstly, the intra prediction mode information of the intra prediction blocks in these adjacent blocks is obtained.

Step 2: Merge adjacent blocks whose adjacent blocks are continuous and whose prediction information is consistent into one adjacent unit.

Specifically, a possible implementation method is to compare whether the intra-frame prediction modes in consecutive adjacent blocks in A1, A2, and A3 are the same, and compare whether the intra-frame prediction modes in consecutive adjacent blocks in B1, B2, and B3 are the same , if two consecutive adjacent blocks have the same intra prediction mode, they are fused into one adjacent unit. According to the method, several adjacent units are obtained, denoted as P1, P2, P3.... For example, a situation may be that the motion vectors in two consecutive adjacent blocks A2 and A3 are the same, which is denoted as P1, and the motion vectors in two consecutive adjacent blocks B1 and B2 are the same, which is denoted as P2. It is also possible that only the motion vectors in the consecutive adjacent blocks of A1 and A2 are the same, which is denoted as P1. It may also be that the motion vectors of A1, A2, and A3 are the same, which is marked as the adjacent unit P1, and the motion vectors of B1, B2, and B3 are the same, which is marked as the adjacent unit P2.

Step 3: Calculate the length of the overlapping boundary between the adjacent unit and the current pixel block, and the obtained length is the length of the adjacent boundary between the adjacent unit and the current pixel block.

Specifically, a possible implementation method is to calculate the length of the coincident boundary between the adjacent unit and the current pixel block according to the coordinate positions of the boundary points P1, P2, P3... and the coordinate positions of the current pixel block boundary points, and the obtained The length is the adjacent boundary length between the adjacent unit and the current pixel block.

The subsequent processing is not included in the patent claims, but it is added to the embodiments in order to make the embodiments clearer and more complete.

Specifically, the adjacent block with the longest boundary adjacent to the current pixel block is used as the reference adjacent unit. Take the intra prediction mode information in the reference neighboring unit as the most probable intra prediction mode of the current block.

Example 5

This example provides a method for deriving the adjacent boundary length of inter-frame prediction blocks in video encoding and decoding, including:

Step 1, at least two adjacent blocks of the current pixel block are obtained.

Specifically, as shown in Figure 1, the adjacent blocks in the spatial domain of the current pixel block include A1, A2, ..., AN, B1, B2, ... BM, among which K blocks are inter-frame prediction blocks, K≥0 . The motion information of the block in the same direction as the reference frame of the current pixel block among the K inter-frame prediction blocks is stored in the input device.

Step 2: Merge adjacent blocks whose adjacent blocks are continuous and whose prediction information is consistent into one adjacent unit.

Specifically, a possible implementation method is to compare whether the motion vector deviations in consecutive adjacent blocks in A1, A2 ... AN, B1, B2, ... BM, are within a small range, if two consecutive The motion vector deviation of adjacent blocks is within a small range, for example, the sum of the absolute values of the differences between the x and y components of the motion vector does not exceed a given threshold, which may be 5. Then they belong to the same adjacent unit. According to the method, several adjacent units are obtained, denoted as P1, P2, P3....

Step 3: Calculate the length of the overlapping boundary between the adjacent unit and the current pixel block, and the obtained length is the length of the adjacent boundary between the adjacent unit and the current pixel block.

Specifically, a possible implementation method is to calculate the length of the coincident boundary between the adjacent unit and the current pixel block according to the coordinate positions of the boundary points P1, P2, P3... and the coordinate positions of the current pixel block boundary points, and the obtained The length is the adjacent boundary length between the adjacent unit and the current pixel block.

The subsequent processing is the same as that described in the last paragraph of Embodiment 1, and will not be repeated here.

Example 6

FIG. 4 is a schematic diagram of a current pixel block and its neighboring blocks in the spatial domain.

This example provides a method for deriving adjacent boundaries of intra prediction blocks in video encoding and decoding, including:

Step 1, at least two adjacent blocks of the current pixel block are obtained.

Specifically, as shown in FIG. 4, there are six adjacent blocks A1, A2, A3, B1, B2, and B3 surrounding the current pixel block. Firstly, the intra prediction mode information of the intra prediction blocks in these adjacent blocks is obtained.

Step 2: Merge adjacent blocks whose adjacent blocks are continuous and whose prediction information is consistent into one adjacent unit.

Specifically, a possible implementation method is to compare whether the angles of intra prediction modes in consecutive adjacent blocks in A1, A2, and A3 are not much different, and compare the angles of intra prediction modes in consecutive adjacent blocks in B1, B2, and B3. Whether the angles of the prediction modes are not much different, if the angles of the intra-frame prediction modes of two consecutive adjacent blocks are not much different, for example, the angle difference of the intra-frame prediction modes is within a given threshold, which may be 10 degrees. Then they belong to the same adjacent unit. According to the method, several adjacent units are obtained, denoted as P1, P2, P3....

Step 3: Calculate the length of the overlapping boundary between the adjacent unit and the current pixel block, and the obtained length is the length of the adjacent boundary between the adjacent unit and the current pixel block.

Specifically, a possible implementation method is to calculate the length of the coincident boundary between the adjacent unit and the current pixel block according to the coordinate positions of the boundary points P1, P2, P3... and the coordinate positions of the current pixel block boundary points, and the obtained The length is the adjacent boundary length between the adjacent unit and the current pixel block.

The subsequent processing is not included in the patent claims, but it is added to the embodiments in order to make the embodiments clearer and more complete.

Specifically, the adjacent block with the longest boundary adjacent to the current pixel block is used as the reference adjacent unit. The intra-frame prediction mode information whose prediction angle in the reference neighboring unit is in the middle is taken out as the most probable intra-frame prediction mode of the current block.

Example 7

FIG. 2 is a schematic diagram of a current pixel block and its adjacent motion vector storage units in the spatial domain.

This example provides a method for deriving adjacent boundaries of inter-frame prediction blocks in video encoding and decoding, including:

Step 1, at least two adjacent blocks existing in the current pixel block are obtained.

Specifically, as shown in FIG. 2, there are motion vector storage units A1-A4, B1-B4, and C1-C3 around the current pixel block. These motion vector storage units are said adjacent blocks. First obtain the motion vectors in these motion vector storage units.

Step 2: Merge adjacent blocks whose adjacent blocks are continuous and whose prediction information is consistent into one adjacent unit.

Specifically, after obtaining a plurality of adjacent blocks in step 1, check all adjacent motion vector storage units A1-A4, B1-B4, and C1-C3 that are continuous and have the same motion vector. The same motion vector storage unit belongs to the same neighboring unit. In Fig. 2, a possible situation is that the prediction information of A4, C3, and B4 are consistent, so they are fused into one adjacent unit.

Step 3: Calculate the length of the overlapping boundary between the adjacent unit and the current pixel block, and the obtained length is the length of the adjacent boundary between the adjacent unit and the current pixel block.

Specifically, a possible implementation method is to calculate the length of the coincident boundary between the adjacent unit and the current pixel block according to the coordinate positions of the boundary points P1, P2, P3... and the coordinate positions of the current pixel block boundary points, and the obtained The length is the adjacent boundary length between the adjacent unit and the current pixel block.

The subsequent processing is the same as that described in the last paragraph of Embodiment 1, and will not be repeated here.

Example 8

This example provides a method for deriving adjacent boundaries of inter-frame prediction blocks in video encoding and decoding, including:

Step 1, at least two adjacent blocks of the current pixel block are obtained.

Specifically, as shown in Figure 1, the adjacent blocks in the spatial domain of the current pixel block include A1, A2, ..., AN, B1, B2, ... BM, among which K blocks are inter-frame prediction blocks, K≥0 . The motion information of the block in the same direction as the reference frame of the current pixel block among the K inter-frame prediction blocks is stored in the input device.

Step 2: Merge adjacent blocks whose adjacent blocks are continuous and whose prediction information is consistent into one adjacent unit.

Specifically, a possible implementation method is to compare whether the motion vector deviations in consecutive adjacent blocks in A1, A2 ... AN, B1, B2, ... BM, are within a small range, such as the motion vector x The sum of the absolute values of the differences with the y component does not exceed a given threshold, which can be 10. , they belong to the same adjacent unit. According to the method, several adjacent units are obtained, denoted as P1, P2, P3....

Step 3: Calculate the length of the overlapping boundary between the adjacent unit and the current pixel block, and the obtained length is the length of the adjacent boundary between the adjacent unit and the current pixel block.

Specifically, a possible implementation method is to calculate the length of the coincident boundary between the adjacent unit and the current pixel block according to the coordinate positions of the boundary points P1, P2, P3... and the coordinate positions of the current pixel block boundary points, and the obtained The length is the adjacent boundary length between the adjacent unit and the current pixel block.

The subsequent processing is the same as that described in the last paragraph of Embodiment 1, and will not be repeated here.

Example 9

This example provides a method for deriving adjacent boundaries of inter-frame prediction blocks in video encoding and decoding, including:

Step 1, at least two adjacent blocks of the current pixel block are obtained.

Specifically, as shown in Figure 1, the adjacent blocks in the spatial domain of the current pixel block include A1, A2, ..., AN, B1, B2, ... BM, among which K blocks are inter-frame prediction blocks, K≥0 . The motion information of the block in the same direction as the reference frame of the current pixel block among the K inter-frame prediction blocks is stored in the input device.

Step 2: Merge adjacent blocks whose adjacent blocks are continuous and whose prediction information is consistent into one adjacent unit.

Specifically, a possible implementation method is to compare whether the motion vector deviations in consecutive adjacent blocks in A1, A2 ... AN, B1, B2, ... BM, are within a small range, such as the motion vector x The larger number in the absolute value of the difference with the y component does not exceed a given threshold, which can be 10, then they belong to the same adjacent unit. A possible situation is that the larger number of the absolute value of the difference between the x and y components of the A1 and A2 motion vectors is 5 and less than 10. Then A1 and A2 are merged into one adjacent unit, and this adjacent unit is marked as P1, and its motion vector is the average value of the motion vectors of A1 and A2. A possible situation is that the greater number in the absolute value of the difference between the x and y components of the motion vectors P1 and A3 is 3, which is less than 10. Then P1 and A3 are fused into one adjacent unit, which is still recorded as P1, and its motion vector is the average value of the original P1 and A3 motion vectors. Such a loop iterative processing. According to the method, several adjacent units are obtained, denoted as P1, P2, P3....

Step 3: Calculate the length of the overlapping boundary between the adjacent unit and the current pixel block, and the obtained length is the length of the adjacent boundary between the adjacent unit and the current pixel block.

Specifically, a possible implementation method is to calculate the length of the coincident boundary between the adjacent unit and the current pixel block according to the coordinate positions of the boundary points P1, P2, P3... and the coordinate positions of the current pixel block boundary points, and the obtained The length is the adjacent boundary length between the adjacent unit and the current pixel block. The subsequent processing is the same as that described in the last paragraph of Embodiment 1, and will not be repeated here.

Example 10

FIG. 3 is a schematic diagram of a current pixel block and its neighboring blocks in the spatial domain.

This example provides a method for deriving adjacent boundaries of inter-frame prediction blocks in video encoding and decoding, including:

Step 1, at least two adjacent blocks of the current pixel block are obtained.

Specifically, as shown in FIG. 3, there are six adjacent blocks A1, A2, A3, B1, B2, and B3 around the current pixel block. Motion vectors in these neighboring blocks are first obtained.

Step 2: Merge adjacent blocks whose adjacent blocks are continuous and whose prediction information is consistent into one adjacent unit.

Specifically, a possible implementation method is to compare whether the scaled motion vectors in consecutive adjacent blocks in A1, A2, and A3 are the same, and compare the scaled motion vectors in consecutive adjacent blocks in B1, B2, and B3 Is it the same. The motion vector scaling method is as follows: the motion vector of the adjacent block is multiplied by the distance between the current block and the reference block of the current block, and then divided by the distance between the adjacent block and the reference block of the adjacent block.

If the scaled motion vectors of any consecutive adjacent blocks are the same, they belong to the same adjacent unit. If the scaled motion vector of a neighboring block is different from the scaled motion vectors of any neighboring blocks to which it is connected, it is considered as a neighboring unit by itself. According to the method, several adjacent units are obtained, denoted as P1, P2, P3.... For example: a situation may be that the scaled motion vectors in two consecutive adjacent blocks A2 and A3 are the same, which is recorded as P1, and the motion vectors in two consecutive adjacent blocks B1 and B2 are the same, which is recorded as P2, A1 is recorded as P3, and B3 is recorded as P4. It may also be that the scaled motion vectors of A1, A2, and A3 are the same, which is marked as the adjacent unit P1, and the scaled motion vectors of B1, B2, and B3 are the same, which is marked as the adjacent unit P2.

Step 3: Calculate the length of the overlapping boundary between the adjacent unit and the current pixel block, and the obtained length is the length of the adjacent boundary between the adjacent unit and the current pixel block.

Specifically, a possible implementation method is to calculate the length of the coincident boundary between the adjacent unit and the current pixel block according to the coordinate positions of the boundary points P1, P2, P3... and the coordinate positions of the current pixel block boundary points, and the obtained The length is the adjacent boundary length between the adjacent unit and the current pixel block. The subsequent processing is the same as that described in the last paragraph of Embodiment 1, and will not be repeated here.

Example 11

FIG. 4 is a schematic diagram of a current pixel block and its neighboring blocks in the spatial domain.

This example provides a method for deriving adjacent boundaries of intra prediction blocks in video encoding and decoding, including:

Step 1, at least two adjacent blocks of the current pixel block are obtained.

Specifically, as shown in FIG. 4, there are six adjacent blocks A1, A2, A3, B1, B2, and B3 around the current pixel block. Firstly, the intra prediction mode information of the intra prediction blocks in these adjacent blocks is obtained.

Step 2: Merge adjacent blocks whose adjacent blocks are continuous and whose prediction information is consistent into one adjacent unit.

Specifically, a possible implementation method is to compare whether the angles of intra prediction modes in consecutive adjacent blocks in A1, A2, and A3 are not much different, and compare the angles of intra prediction modes in consecutive adjacent blocks in B1, B2, and B3. Whether the difference between prediction mode angles is within 30 degrees. If the difference between the intra prediction mode angles of two consecutive adjacent blocks is not large, they belong to the same adjacent unit. According to the method, several adjacent units are obtained, denoted as P1, P2, P3....

Step 3: Calculate the length of the overlapping boundary between the adjacent unit and the current pixel block, and the obtained length is the length of the adjacent boundary between the adjacent unit and the current pixel block.

Specifically, a possible implementation method is to calculate the length of the coincident boundary between the adjacent unit and the current pixel block according to the coordinate positions of the boundary points P1, P2, P3... and the coordinate positions of the current pixel block boundary points, and the obtained The length is the adjacent boundary length between the adjacent unit and the current pixel block.

Example 12

FIG. 5 is a schematic structural diagram of an apparatus for deriving adjacent boundary lengths in video coding and decoding according to an embodiment of the present invention. As shown in FIG. 5 , in this embodiment, the apparatus for deriving motion vectors in video coding and decoding includes an input module 310 , a fusion module 320 , and an adjacent boundary length calculation module 330 for calculating adjacent boundary lengths.

The input module 310 provided in this embodiment is specifically used for: as shown in FIG. 1 , the adjacent blocks in the spatial domain of the current pixel block include A1, A2, ..., AN, B1, B2, ... BM, of which there are K The block is an inter prediction block, K≥0. The motion information of the block in the same direction as the reference frame of the current pixel block among the K inter-frame prediction blocks is stored in the input device.

The adjacent block merging module 320 provided in this embodiment is specifically used for: merging adjacent blocks whose prediction information is consistent in consecutive adjacent blocks into one adjacent unit. Specifically, as shown in Figure 1, compare A1, A2 ... AN, B1, B2, ... BM, whether the motion vectors in consecutive adjacent blocks are the same, if the motion vectors of two consecutive adjacent blocks are the same, then They belong to the same adjacent unit. According to the method, several adjacent units are obtained, denoted as P1, P2, P3....

The reference adjacent block selection unit output unit 330 provided in this embodiment is specifically used for:

Calculate the length of the overlapping boundary between the adjacent unit and the current pixel block, and the resulting length is the length of the adjacent boundary between the adjacent unit and the current pixel block.

Specifically, according to the obtained boundary point coordinates of the adjacent unit and the boundary point coordinates of the current pixel block, the coincident boundary between the adjacent unit and the current pixel block is calculated, which is the adjacent boundary length.

Those of ordinary skill in the art can understand that all or part of the steps for realizing the above-mentioned method embodiments can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, the It includes the steps of the above method embodiments; and the aforementioned storage medium includes: ROM, RAM, magnetic disk or optical disk and other various media that can store program codes.

Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: it can still be Modifications are made to the technical solutions described in the foregoing embodiments, or equivalent replacements are made to some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions depart from the scope of the technical solutions of the embodiments of the present invention.

Claims (8)

1. an adjacent boundary length deriving method is characterized in that, may further comprise the steps:

Obtain at least two adjacent blocks of current pixel piece;

With adjacent block for continuously and the adjacent block of its an information of forecasting unanimity adjacent cells that permeates;

Calculate the length of adjacent cells and current pixel piece coincidence boundary, Len got is adjacent cells and current pixel piece adjacent boundary length.

2. adjacent boundary length deriving method as claimed in claim 1 is characterized in that, the information of forecasting in the described continuous adjacent piece is consistent to refer to that the same or deviation of the movable information of inter prediction piece is in a little scope.

3. adjacent boundary length deriving method as claimed in claim 1 is characterized in that, the information of forecasting in the described continuous adjacent piece is consistent to refer to that the same or deviation of directivity of the intra prediction mode of intra-frame prediction block is in a little scope.

4. adjacent boundary length deriving method as claimed in claim 2, it is characterized in that the movable information of described inter prediction piece is: the value that the motion vector of the motion vector of inter prediction piece or inter prediction piece obtains behind the convergent-divergent according to the frame pitch between described inter prediction piece and the described inter prediction piece reference block.

5. an adjacent boundary length let-off gear(stand) is characterized in that, comprising:

The input module that is used at least two adjacent blocks of acquisition current pixel piece;

Be used for adjacent block for continuously and the permeate adjacent cells Fusion Module of an adjacent cells of the adjacent block of its information of forecasting unanimity;

Be used for calculating the length of adjacent cells and current pixel piece coincidence boundary, Len got is the adjacent boundary length computation module of adjacent cells and current pixel piece adjacent boundary length.

6. adjacent boundary length let-off gear(stand) as claimed in claim 5 is characterized in that, the information of forecasting in the described continuous adjacent piece is consistent to refer to that the same or deviation of the motion vector of inter prediction piece is in a little scope.

7. adjacent boundary length let-off gear(stand) as claimed in claim 5 is characterized in that, the information of forecasting in the described continuous adjacent piece is consistent to refer to that the same or deviation of directivity of the intra prediction mode of intra-frame prediction block is in a little scope.

8. adjacent boundary length let-off gear(stand) as claimed in claim 6, it is characterized in that the movable information of described inter prediction piece is: the value that the motion vector of the motion vector of inter prediction piece or inter prediction piece obtains behind the convergent-divergent according to the frame pitch between described inter prediction piece and the described inter prediction piece reference block.

Images