RU2776356C1 - Method and device for encoding and decoding, equipment on the encoding side and equipment on the decoding side - Google Patents

Abstract

FIELD: predictive coding and decoding

SUBSTANCE: invention relates to the field of predictive coding and decoding. A method for encoding and decoding is proposed and includes the following steps: if the characteristic information of the current block meets a special condition, the first reference block corresponding to the current block is determined based on the first initial motion vector of the current block, and the second reference block corresponding to the current block is determined based on the second the initial motion vector of the current block; adjusting the first and second source motion vectors according to the first pixel value of the first reference block and the second pixel value of the second reference block to obtain a first target motion vector corresponding to the first source motion vector and a second target motion vector corresponding to the second source motion vector; encoding or decoding the current block based on the first and second target motion vectors.

EFFECT: improvement of coding performance, in particular, prediction quality, prediction error.

15 cl, 11 dwg

Description

The technical field to which the invention belongs

[01] The present application relates to encoding and decoding technologies, in particular, to an encoding and decoding method and equipment, an encoding-side apparatus, and a decoding-side apparatus.

State of the art

[02] To save space, transmission of video images is carried out after encoding them, and the method of completely encoding video data may include the processes of prediction, transformation, quantization, entropy coding, filtering, and the like. Predictive coding includes inter-frame coding and intra-frame coding, wherein inter-frame coding predicts the pixels of the current image according to time domain correlation of video data to effectively eliminate video data redundancy in time domain by pixels of an adjacent encoded image.

[03] In the process of inter-frame coding, it is possible to use a motion vector (VD, English Motion Vector (MV)), reflecting the relative offset between the block of the current image of the video image of the current frame and the block of the reference video image of the reference frame. For example, if there is a strong correlation in the time domain between the video image A of the current frame and the video image B of the reference frame, if the image block A1 (current image block) of the video image A is to be transmitted, motion search can be performed in the video image B to detect the image block B1 (i.e., . of the reference picture block) best matching the picture block A1, and to determine the relative offset between the picture block A1 and the picture block B1, wherein the relative shift is also the motion vector of the picture block A1.

[04] The encoding side may send a motion vector to the decoding side instead of sending the image block A1 to the decoding side. The decoding side may obtain the picture block A1 based on the motion vector and the picture block B1. Obviously, the above solution can save a large number of bits, since the number of bits occupied by the motion vector is less than the number of bits occupied by the image block A1.

[05] According to the known method, if the current block is a unidirectional block, after the motion vector of the current block (hereinafter referred to as "original motion vector") is obtained, the original motion vector can be corrected and encoding/decoding based on the corrected motion vector which improves coding performance. Meanwhile, if the current block is a bidirectional block, there is currently no rational solution to the problem of how to correct the first original motion vector and the second original motion vector after obtaining the first and second original motion vectors of the current block. That is, in the case of a bidirectional block, problems such as poor prediction quality, prediction error, and the like are possible, leading to poor coding performance.

The essence of the invention

[06] The present application proposes a method, equipment and apparatus for encoding and decoding to improve encoding performance.

[07] The present application provides a method for encoding and decoding, including the steps of: if the characteristic information of the current block meets a specific condition, determining the first reference block corresponding to the current block based on the first original motion vector of the current block, and determining the second a reference block corresponding to the current block based on the second original motion vector of the current block; adjusting the first and second source motion vectors according to the first pixel value of the first reference block and the second pixel value of the second reference block to obtain a first target motion vector corresponding to the first source motion vector and a second target motion vector corresponding to the second source motion vector; encoding or decoding the current block based on the first and second target motion vectors.

[08] The present application provides encoding and decoding equipment, comprising: a determination module configured to, if the characteristic information of the current block meets a specific condition, determine the first reference block corresponding to the current block based on the first original motion vector of the current block, and determine a second reference block corresponding to the current block, based on the second original motion vector of the current block; a processing module configured to correct the first and second source motion vectors according to the first pixel value of the first reference block and the second pixel value of the second reference block to obtain a first target motion vector corresponding to the first source motion vector and a second target motion vector corresponding to the second source vector movement; and an encoding and decoding module, configured to encode or decode the current block based on the first and second target motion vectors.

[09] The present application provides an encoding-side apparatus comprising a processor and a computer-readable storage medium, the computer-readable storage medium having computer-executable instructions stored thereon which, when executed by the processor, initiate a set of operations in which: if the characteristic information of the current block meets the specific condition, determining a first reference block corresponding to the current block based on the first source motion vector of the current block, and determining a second reference block corresponding to the current block based on the second source motion vector of the current block; adjusting the first and second source motion vectors according to the first pixel value of the first reference block and the second pixel value of the second reference block to obtain a first target motion vector corresponding to the first source motion vector and a second target motion vector corresponding to the second source motion vector; and encoding the current block based on the first and second target motion vectors.

[10] The present application provides a device on the decoding side, comprising a processor and a computer-readable storage medium, and the computer-readable storage medium contains computer-executable instructions stored thereon, which, when executed by the processor, initiate the execution of a set of operations in which: if the characteristic information of the current block meets the specific condition, determining a first reference block corresponding to the current block based on the first source motion vector of the current block, and determining a second reference block corresponding to the current block based on the second source motion vector of the current block; adjusting the first and second source motion vectors according to the first pixel value of the first reference block and the second pixel value of the second reference block to obtain a first target motion vector corresponding to the first source motion vector and a second target motion vector corresponding to the second source motion vector; and decoding the current block based on the first and second target motion vectors.

[11] The present application proposes a decoding method, including the steps of: when the current block allows the motion vector refinement mode, determine the prediction values of the sub-blocks of the current block, and determine the prediction value of the current block based on the prediction values of the sub-blocks of the current block, and for of each subblock of the current block, determining the subblock prediction value includes: obtaining motion information of the current block, comprising a first original motion vector and a first reference frame, a second original motion vector and a second reference frame; determining a first reference block corresponding to the subblock based on the first source motion vector of the current block, and determining a second reference block corresponding to the subblock based on the second source motion vector of the current block; obtaining the optimal motion vector based on the pixel values of the first reference block and the pixel values of the second reference block, wherein obtaining the optimal motion vector includes: taking the original motion vector as the center, selecting the motion vectors from the motion vectors containing the original motion vector and the motion vectors , surrounding the original motion vector, and define the selected motion vectors as potential motion vectors; wherein the original motion vector is a first original motion vector or a second original motion vector; obtaining a cost value corresponding to the original motion vector and cost values corresponding to the potential motion vectors based on the pixel values of the first reference block and the pixel values of the second reference block; selecting a motion vector from the original motion vector and the potential motion vectors as the optimal motion vector based on the cost value corresponding to the original motion vector and the cost values corresponding to the potential motion vectors; adjusting the first source motion vector and the second source motion vector according to the optimal motion vector to obtain a first target motion vector corresponding to the first source motion vector and a second target motion vector corresponding to the second source motion vector; determining a subblock prediction value based on the first target motion vector and the second target motion vector; wherein determining the sub-block prediction value based on the first target motion vector and the second target motion vector includes: determining a third reference block corresponding to the sub-block based on the first target motion vector, and determining a fourth reference block corresponding to the sub-block based on the second target motion vector movement; weighing the pixel values of the third reference block and the pixel values of the fourth reference block to obtain a sub-block prediction value.

[12] As an example, the step of taking the original motion vector as the center, selecting motion vectors from motion vectors containing the original motion vector and motion vectors surrounding the original motion vector, and determining the selected motion vectors as potential motion vectors, includes: taking the original motion vector as the center, searching for 25 motion vectors containing the original motion vector in a search range of 2, and determining 25 motion vectors as candidate motion vectors; wherein the search sequence of 25 motion vectors is: { Mv(-2, -2), Mv(-1, -2), Mv(0, -2), Mv(1, -2), Mv(2, - 2), Mv(-2, -1), Mv(-1, -1), Mv(0, -1), Mv(1, -1), Mv(2, -1), Mv(-2, 0), Mv(-1, 0), Mv(0, 0), Mv(1, 0), Mv(2, 0), Mv(-2, 1), Mv(-1, 1), Mv( 0, 1), Mv(1, 1), Mv(2, 1), Mv(-2, 2), Mv(-1, 2), Mv(0, 2), Mv(1, 2), Mv (2, 2) }.

[13] As an example, the step of obtaining the cost value corresponding to the original motion vector and the cost values corresponding to the potential motion vectors based on the pixel values of the first reference block and the pixel values of the second reference block includes the following: obtaining the cost values , corresponding to the potential motion vectors, includes: determining the first reference sub-blocks corresponding to the potential motion vectors based on the first reference block, and determining the second reference sub-blocks corresponding to the potential motion vectors based on the second reference block; calculate the sum of the absolute differences of the sampled values of the pixel dots in the first reference sub-blocks and the second reference sub-blocks by using a 2-fold vertical downsampling operation; determining cost values corresponding to potential motion vectors based on the obtained sum of absolute differences; obtaining a cost value corresponding to the original motion vector includes: determining a first reference subblock corresponding to the original motion vector based on the first reference block, and determining a second reference subblock corresponding to the original motion vector based on the second reference block; calculate the sum of the absolute differences of the sampled values of the pixel dots in the first reference subblock and the second reference subblock by using a 2-fold vertical downsampling operation; determine the value of the cost corresponding to the original motion vector, based on the obtained sum of absolute differences.

[14] As an example, the step of adjusting the first source motion vector and the second source motion vector according to the optimal motion vector to obtain the first target motion vector corresponding to the first source motion vector and the second target motion vector corresponding to the second source motion vector, includes: determining a first integer pixel motion vector correction value and a second integer pixel motion vector correction value based on the optimal motion vector; correcting the first source motion vector according to the first integer pixel motion vector correction value to obtain a first target motion vector corresponding to the first source motion vector; and adjusting the second source motion vector according to the second integer pixel motion vector correction value to obtain a second target motion vector corresponding to the second source motion vector.

[15] As an example, the step of determining the first integer pixel motion vector correction value and the second integer pixel motion vector correction value based on the optimal motion vector includes: determining the first integer pixel motion vector correction value based on the optimal motion vector, the first the integer value of the motion vector correction is equal to the difference between the optimal motion vector and the original motion vector; determining a second integer pixel motion vector correction value based on the first integer pixel motion vector correction value; wherein the second integer pixel motion vector correction value and the first integer pixel motion vector correction value are opposite to each other.

[16] As an example, the step of correcting the first source motion vector according to the first integer-pixel motion vector correction value to obtain the first target motion vector corresponding to the first source motion vector; and adjusting the second source motion vector according to the second integer pixel motion vector correction value to obtain a second target motion vector corresponding to the second source motion vector, including: the first target motion vector is equal to the sum of the first source motion vector and the first integer pixel motion vector correction value; the second target motion vector is equal to the sum of the second source motion vector and the second integer pixel motion vector correction value.

[17] As an example, the step of adjusting the first source motion vector and the second source motion vector according to the optimal motion vector to obtain the first target motion vector corresponding to the first source motion vector and the second target motion vector corresponding to the second source motion vector, includes: determining a first integer pixel motion vector correction value, a first subpixel motion vector correction value, a second integer pixel motion vector correction value, and a second subpixel motion vector correction value based on the optimal motion vector; correcting the first source motion vector according to the first integer motion vector correction value and the first subpixel motion vector correction value to obtain the first target motion vector of the subblock; and correcting the second source motion vector according to the second integer motion vector correction value and the second subpixel motion vector correction value to obtain a second target motion vector of the subblock.

[18] As an example, the step of adjusting the first source motion vector and the second source motion vector according to the optimal motion vector to obtain the first target motion vector corresponding to the first source motion vector and the second target motion vector corresponding to the second source motion vector, includes: determining a first integer pixel motion vector correction value and a first subpixel motion vector correction value based on the optimal motion vector; obtaining an optimal offset motion vector according to the first integer motion vector correction value and the first subpixel motion vector correction value; adjusting the first source motion vector based on the optimal offset motion vector to obtain a first target motion vector; adjusting the second source motion vector based on the opposite value of the optimal offset motion vector to obtain the second target motion vector

[19] As an example, the step of determining the first integer pixel motion vector correction value and the first subpixel motion vector correction value based on the optimal motion vector includes: determining the first integer pixel motion vector correction value based on the optimal motion vector, the first the integer value of the motion vector correction is equal to the difference between the optimal motion vector and the original motion vector; determining a first sub-pixel motion vector correction value based on a cost value corresponding to the optimal motion vector and cost values corresponding to edge motion vectors corresponding to the optimal motion vector.

[20] As an example, the step of determining the first sub-pixel motion vector correction value based on the cost value corresponding to the optimal motion vector and the cost values corresponding to the edge motion vectors corresponding to the optimal motion vector includes: sequentially determining the cost values 5 whole-pixel motion vectors, for which the optimal motion vector is taken as the center; wherein the 5 integer motion vectors are 5 edge motion vectors obtained by shifting horizontally left, horizontally right, vertically up, and vertically down, respectively, when the optimal motion vector is taken as the center.

[21] As an example, the first target motion vector is equal to the sum of the first source motion vector, the first integer-pixel motion vector correction value, and the first sub-pixel motion vector correction value; the second target motion vector is equal to the sum of the second source motion vector, the second integer-pixel motion vector correction value, and the second sub-pixel motion vector correction value; wherein the second integer pixel motion vector correction value and the first integer pixel motion vector correction value are opposite to each other; the second sub-pixel motion vector correction value and the first sub-pixel motion vector correction value are opposite to each other.

[22] As an example, the optimal motion vector is the lower cost motion vector selected from the original motion vector and potential motion vectors.

[23] As an example, when the absolute value of the first integer pixel motion vector correction value in the vertical and/or horizontal direction is equal to the predetermined value, adjust the first source motion vector according to the first integer pixel motion vector correction value to obtain the first target motion vector corresponding to the first source vector movement.

[24] As an example, when the absolute value of the first integer pixel motion vector correction value in the vertical and/or horizontal direction is not equal to the predetermined value, the first original motion vector is corrected according to the first integer pixel motion vector correction value and the first subpixel motion vector correction value to obtain the first a target motion vector corresponding to the first source motion vector.

[25] As an example, when the cost value corresponding to the original motion vector is greater than or equal to a predetermined threshold, the step of selecting a motion vector from the original motion vector and potential motion vectors as the optimal motion vector based on the cost value corresponding to the original motion vector is performed, and cost values corresponding to potential motion vectors; otherwise, this step is not performed, and the predetermined threshold is determined based on the width and height of the sub-block.

[26] As an example, when the current block contains one subblock, the specified subblock is the current block.

[27] As an example, the first reference block and the second reference block are luminance prediction blocks determined based on the luminance component.

[28] As an example, both the pixel values of the first reference block and the pixel values of the second reference block are obtained by bilinear interpolation.

[29] As an example, the step of determining the first reference block corresponding to the subblock based on the first original motion vector of the current block and determining the second reference block corresponding to the subblock based on the second original motion vector of the current block includes: determining the first a reference block by bilinear interpolation based on the first original motion vector and the first reference frame of the current block; determining a second reference block by bilinear interpolation based on the second original motion vector and the second reference frame of the current block; wherein the size of the first reference block is equal to the size of the second reference block, wherein the width value of the first reference block is determined based on the sub-block width value and the search range, the height value of the first reference block is determined based on the sub-block height value and the search range.

[30] As an example, the step of determining the first reference block corresponding to the subblock based on the first original motion vector of the current block, and determining the second reference block corresponding to the subblock based on the second original motion vector of the current block, includes: obtaining the first integer pixel block from the first reference frame of the current block based on the first original motion vector, bilinearly interpolating the first integer pixel block to obtain the first reference block; obtaining a second integer pixel block from the second reference frame of the current block based on the second original motion vector, performing bilinear interpolation of the second integer pixel block to obtain a second reference block; if W is the sub-box width, H is the sub-box height, and SR is the search range, then the widths of the first ref-box and the second ref-box are both W+2*SR, the heights of the first ref-box and the second ref-box are both H +2*SR, with the value of SR being 2.

[31] As an example, the pixel values of the third reference block and the pixel values of the fourth reference block, when weighted, have the same weight.

[32] As an example, the pixel values of the third reference block and the pixel values of the fourth reference block are obtained based on interpolation with an 8-tap interpolation filter.

[33] As an example, determining a third reference block corresponding to a subblock based on a first target motion vector and determining a fourth reference block corresponding to a subblock based on a second target motion vector includes: determining a fifth reference block , corresponding to the sub-block, from the first reference frame of the fluid block based on the first target motion vector, and interpolating the pixel values in the fifth reference block based on the 8-tap interpolation filter to obtain the third reference block; determining a sixth reference block corresponding to the sub-block from the second reference frame based on the second target motion vector, and interpolating pixel values in the sixth reference block based on the 8-tap interpolation filter to obtain a fourth reference block; wherein the size of the fifth reference block is greater than the size of the sub-block, the size of the sixth reference block is greater than the size of the sub-block, said dimension including the width and the height.

[34] As an example, the sub-block contains a luminance component, and performing motion compensation with sub-block weighting according to the first target motion vector and the second target motion vector to obtain a weighted sub-block prediction value, includes: for the luminance component of the sub-block, performing motion compensation with sub-block weighting according to the first target motion vector and the second target motion vector to obtain a weighted sub-block brightness prediction value.

[35] As an example, the sub-block contains a chrominance component, and performing motion compensation with sub-block weighting according to the first target motion vector and the second target motion vector to obtain the sub-block prediction weighted value includes: for the chrominance component of the sub-block, performing motion compensation with sub-block weighting according to to the first target motion vector and the second target motion vector to obtain a weighted sub-block chrominance prediction value.

[36] As an example, when the current block allows the motion vector refinement mode, the characteristic information of the current block corresponds to specific conditions, the specific conditions include the motion information prediction mode corresponding to the current block; a motion information attribute corresponding to the current block and size information of the current block; wherein, when the current block allows the motion vector refinement mode, the conditions to which the current block corresponds include at least the following: the merge mode for encoding motion information difference values (MMVD mode) is not applied to the current block, the merge mode is applied to the current block merge, the size of the current block is within the specified range, the motion information of the current block includes motion information in two different directions, and the distances between the current frame and each of the two reference frames corresponding to the motion information in two different directions are the same .

[37] As an example, if the current block does not meet any of the following conditions, the current block does not allow the motion vector refinement mode: the merge mode for encoding motion information difference values (MMVD mode) is not applied to the current block, the merge mode is applied to the current block merge mode, the size of the current block is within the specified range, the motion information of the current block includes motion information in two different directions, and the distances between the current frame and each of the two reference frames corresponding to the motion information in two different directions are the same.

[38] As an example, the distance between the current frame of the current block and the first key frame is equal to the distance between the current frame of the current block and the second key frame.

[39] As an example, when the current block contains one subblock, the specified subblock is the current block, and the first source motion vector and the second source motion vector are the source motion vectors of the current block in two different directions.

[40] As an example, when the current block contains a plurality of sub-blocks, each of the plurality of sub-blocks, the first original motion vector and the second original motion vector have common original motion vectors of the current block in two different directions.

[41] According to another aspect, the present application proposes an encoding method, including the steps of: when the current block allows the motion vector refinement mode, determine the prediction values of the sub-blocks of the current block, and determine the prediction value of the current block based on the prediction values of the sub-blocks of the current block, wherein for each subblock of the current block, determining the subblock prediction value includes: obtaining motion information of the current block, comprising a first original motion vector and a first reference frame, a second original motion vector and a second reference frame; determining a first reference block corresponding to the subblock based on the first source motion vector of the current block, and determining a second reference block corresponding to the subblock based on the second source motion vector of the current block; obtaining the optimal motion vector based on the pixel values of the first reference block and the pixel values of the second reference block, wherein obtaining the optimal motion vector includes: taking the original motion vector as the center, selecting the motion vectors from the motion vectors containing the original motion vector and the motion vectors , surrounding the original motion vector, and define the selected motion vectors as potential motion vectors; wherein the original motion vector is a first original motion vector or a second original motion vector; obtaining a cost value corresponding to the original motion vector and cost values corresponding to the potential motion vectors based on the pixel values of the first reference block and the pixel values of the second reference block; selecting a motion vector from the original motion vector and the potential motion vectors as the optimal motion vector based on the cost value corresponding to the original motion vector and the cost values corresponding to the potential motion vectors; adjusting the first source motion vector and the second source motion vector according to the optimal motion vector to obtain a first target motion vector corresponding to the first source motion vector and a second target motion vector corresponding to the second source motion vector; determining a subblock prediction value based on the first target motion vector and the second target motion vector; wherein determining the sub-block prediction value based on the first target motion vector and the second target motion vector includes: determining a third reference block corresponding to the sub-block based on the first target motion vector, and determining a fourth reference block corresponding to the sub-block based on the second target motion vector movement; weighing the pixel values of the third reference block and the pixel values of the fourth reference block to obtain a sub-block prediction value.

[42] According to another aspect, the present application provides decoding equipment, comprising: a processor and a computer-readable storage medium, the computer-readable storage medium having computer-executable instructions stored thereon that can be executed by the processor; wherein the processor is configured to be used to execute machine-executable instructions for implementing the above method.

[43] According to another aspect, the present application provides encoding equipment, comprising: a processor and a computer-readable storage medium, the computer-readable storage medium having computer-executable instructions stored thereon, which can be executed by the processor; wherein the processor is configured to be used to execute computer-executable instructions to implement the steps where: when the current block allows the motion vector refinement mode, determine the prediction values of the sub-blocks of the current block, and determine the prediction value of the current block based on the prediction values of the sub-blocks of the current block, and for of each subblock of the current block, determining the subblock prediction value includes: obtaining motion information of the current block, comprising a first original motion vector and a first reference frame, a second original motion vector and a second reference frame; determining a first reference block corresponding to the subblock based on the first source motion vector of the current block, and determining a second reference block corresponding to the subblock based on the second source motion vector of the current block; obtaining the optimal motion vector based on the pixel values of the first reference block and the pixel values of the second reference block, wherein obtaining the optimal motion vector includes: taking the original motion vector as the center, selecting the motion vectors from the motion vectors containing the original motion vector and the motion vectors , surrounding the original motion vector, and define the selected motion vectors as potential motion vectors; wherein the original motion vector is a first original motion vector or a second original motion vector; obtaining a cost value corresponding to the original motion vector and cost values corresponding to the potential motion vectors based on the pixel values of the first reference block and the pixel values of the second reference block; selecting a motion vector from the original motion vector and the potential motion vectors as the optimal motion vector based on the cost value corresponding to the original motion vector and the cost values corresponding to the potential motion vectors; adjusting the first source motion vector and the second source motion vector according to the optimal motion vector to obtain a first target motion vector corresponding to the first source motion vector and a second target motion vector corresponding to the second source motion vector; determining a subblock prediction value based on the first target motion vector and the second target motion vector; wherein determining the sub-block prediction value based on the first target motion vector and the second target motion vector includes: determining a third reference block corresponding to the sub-block based on the first target motion vector, and determining a fourth reference block corresponding to the sub-block based on the second target motion vector movement; weighing the pixel values of the third reference block and the pixel values of the fourth reference block to obtain a sub-block prediction value.

[44] According to another aspect, the present application proposes a decoding device capable of executing the above method.

[45] According to another aspect, the present application provides an encoder capable of executing the steps of: when the current block allows the motion vector refinement mode, determine the prediction values of the sub-blocks of the current block, and determine the prediction value of the current block based on the prediction values of the sub-blocks of the current block, wherein, for each subblock of the current block, determining the subblock prediction value includes: obtaining motion information of the current block comprising a first source motion vector and a first reference frame, a second source motion vector and a second reference frame; determining a first reference block corresponding to the subblock based on the first source motion vector of the current block, and determining a second reference block corresponding to the subblock based on the second source motion vector of the current block; obtaining the optimal motion vector based on the pixel values of the first reference block and the pixel values of the second reference block, wherein obtaining the optimal motion vector includes: taking the original motion vector as the center, selecting the motion vectors from the motion vectors containing the original motion vector and the motion vectors , surrounding the original motion vector, and define the selected motion vectors as potential motion vectors; wherein the original motion vector is a first original motion vector or a second original motion vector; obtaining a cost value corresponding to the original motion vector and cost values corresponding to the potential motion vectors based on the pixel values of the first reference block and the pixel values of the second reference block; selecting a motion vector from the original motion vector and the potential motion vectors as the optimal motion vector based on the cost value corresponding to the original motion vector and the cost values corresponding to the potential motion vectors; adjusting the first source motion vector and the second source motion vector according to the optimal motion vector to obtain a first target motion vector corresponding to the first source motion vector and a second target motion vector corresponding to the second source motion vector; determining a subblock prediction value based on the first target motion vector and the second target motion vector; wherein determining the sub-block prediction value based on the first target motion vector and the second target motion vector includes: determining a third reference block corresponding to the sub-block based on the first target motion vector, and determining a fourth reference block corresponding to the sub-block based on the second target motion vector movement; weighing the pixel values of the third reference block and the pixel values of the fourth reference block to obtain a sub-block prediction value.

[46] According to another aspect, the present application provides a durable computer-readable storage medium containing computer-executable instructions stored thereon, which can be executed by a processor, and the processor is configured to be used to execute computer-executable instructions for implementing the above method.

[47] From the above technical solutions it is obvious that in the embodiment of the present application, the first and second target motion vectors of the current block can be determined based on the first and second source motion vectors, while the current block can be encoded or decoded based on the first and second target vectors motion instead of encoding or decoding the current block based on the first and second original motion vectors directly, to solve problems such as poor prediction quality, prediction error, and the like, and improve encoding performance and increase encoding efficiency.

Brief description of the drawings

[48] In order to give a clearer idea of the technical solution in the embodiment of the present application or the known solution, the drawings for use in these embodiments and the drawings of the known solution will be briefly described below. Of course, the drawings below are intended only for some embodiments of the present application; however, those skilled in the art will be able to obtain other drawings based on these drawings.

[49] FIG. 1A is a schematic diagram of an interpolation according to an embodiment of the present application;

[50] FIG. 1B is a schematic diagram of a video coding infrastructure according to an embodiment of the present application;

[51] FIG. 2 is a flowchart of an encoding and decoding method according to an embodiment of the present application;

[52] FIG. 3 is a flowchart of an encoding and decoding method according to an embodiment of the present application;

[53] FIG. 4 is a flowchart of an encoding and decoding method according to an embodiment of the present application;

[54] FIG. 5 is a schematic diagram of reference pixels according to an embodiment of the present application;

[55] FIG. 6 is a schematic diagram of a motion vector iteration according to an embodiment of the present application;

[56] FIG. 7A-7E are schematic diagrams of a sequence of potential points according to an embodiment of the present application;

[57] FIG. 8 is a schematic diagram of an anchor in the spatial domain of a current block according to an embodiment of the present application;

[58] FIG. 9 is a block diagram of the encoding and decoding equipment of an embodiment of the present application;

[59] FIG. 10 is a hardware block diagram of a device on the decoding side of an embodiment of the present application;

[60] FIG. 11 is a hardware block diagram of an encoding-side device according to an embodiment of the present application.

Implementation of the invention

[61] The terms used in the embodiments of the present application are for disclosure purposes only and are not intended to limit the present application. In it, unless otherwise indicated and / or otherwise clearly follows from the context, indefinite articles (English "a" or "an") mean "at least one", and the definite article (English "the") means " specified at least one. It should be understood that the expression "and/or" in this application is used to denote all possible combinations of one or more of the listed positions to which it refers.

[62] It should be understood that, despite the possibility of using the concepts of "first", "second", "third", etc. regarding the various elements, these elements should not be limited by these concepts. They serve solely to distinguish between elements of different types. For example, the "first" element may also be referred to as the "second" element, and the "second" element may similarly be referred to as the "first" element without departing from the scope of the present application. The word "if" can mean "when ..." or "when ...", "as a reaction to the definition", depending on the context.

[63] In an embodiment of the present application, an encoding and decoding method is proposed, with which the following concepts can be associated:

[64] Intra-picture prediction and inter-picture prediction techniques: The intra-picture prediction technique is to apply the pixels of the encoded block of the current image to predict the current pixels according to the spatial correlation of the video data to effectively eliminate the redundancy of the video data in the spatial domain. The intra-frame prediction technique provides for setting a plurality of prediction modes, each of which corresponds to one texture direction (except for the direct coding mode), while the prediction signal of the current block is generated based on the value of the boundary reconstructed pixel of the block adjacent to the current block in the prediction direction. For example, if the image texture is horizontally oriented, the image information is better predicted by selecting the horizontal prediction mode. Since a video sequence usually has a stronger time domain correlation, an inter-picture prediction technique should use pixels of an adjacent encoded image to predict the current pixels of the current image according to the time domain correlation of the video data to effectively eliminate video data redundancy in the time domain. The section on interframe prediction of the main video data coding standard provides for the use of block motion compensation technology, the basic principle of which is to obtain an optimally matching block in a previously encoded image for each block of pixels in the current image, called Motion Estimation (ME) ).

[65] Motion vector (VD): In inter-frame coding, motion vectors serve to indicate the relative offset between the currently encoded block and the optimally matched block in the corresponding reference picture. Each divided block has a corresponding motion vector, which is transmitted to the decoding side, and a significant number of bits are consumed for independent encoding and transmission of the motion vector of each block, in particular, divided into small blocks. To reduce the number of bits for motion vector coding, spatial correlation between adjacent image blocks is used to predict the motion vector of the encoded current block according to the motion vector of the adjacent encoded block, followed by transmission of the predicted difference. This effectively reduces the number of bits representing the motion vector. In the process of encoding the motion vector of the current block, the motion vector of the current block is predicted based on the motion vector of the adjacent encoded block, followed by encoding the motion vector difference (MVD, eng. Motion Vector Difference (MVD)) between the predicted motion vector (PVD, Motion Vector Prediction (MVP) )) and the actual estimated value of the motion vector, which significantly reduces the number of encoded TD bits.

[66] Motion information: since the motion vector indicates the offset between the current picture block and a certain reference picture block, in order to collect accurate information pointing to this picture block, in addition to the motion vector, reference frame picture index information is needed to indicate the applied picture reference frame. The video coding technology provides for the possibility of creating a common key frame picture list for the current frame picture, wherein the key frame picture index information indicates the key frame picture in the key frame picture list received by the current picture block. In addition, many coding techniques also allow multiple reference picture lists, so it is also possible to use an index value, which may be referred to as "reference direction", to indicate which reference picture list to use. In video coding technology, motion-related information such as motion vector, reference frame index, and reference direction may be collectively referred to as "motion information".

[67] Prediction signal: The prediction signal is a pixel value obtained based on an already encoded and decoded pixel, with a residual obtained based on the difference between the original pixel and the predicted pixel, and then transforming and quantizing the residual and encoding the coefficients. In particular, the inter-picture prediction signal refers to a pixel value obtained based on a reference frame (a frame of reconstructed pixels) of the current block, whereby the final predicted pixel is obtained by interpolation due to the discrete arrangement of pixels. The closer the predicted pixel is to the original pixel, the smaller the residual energy obtained by subtracting the predicted pixel from the original pixel, and the better the encoding compression performance.

[68] Interpolation: If the accuracy of the current VD is not integer pixel, the existing pixel value cannot be directly copied from the corresponding reference frame, and the required pixel value can only be obtained by interpolation. FIG. 1A shows that if a Y1/2 pixel value with a 1/2 pixel offset is to be obtained, it is obtained by interpolating the X values of the surrounding existing pixels. When using an interpolation filter with N taps, interpolation is performed by interpolating N surrounding integer pixels.

[69] Motion compensation: motion compensation is the process of obtaining all pixel values of the current block by interpolation or copying.

[70] Video coding infrastructure: as disclosed in FIG. 1B, the encoding-side processing sequence of an embodiment of the present application can be implemented with a video encoding infrastructure, the circuit diagram of the video decoding infrastructure is similar to that of FIG. embodiment of the present application can be implemented using a video decoding infrastructure. In particular, the video coding infrastructure and the video decoding infrastructure include modules such as intra-frame prediction, motion estimation/motion compensation, reference picture buffering, in-loop filtering, reconstruction, transform, quantization, inverse transform, inverse quantization, entropy coding, etc. P. The processing sequence on the encoding side can be implemented by interaction between these modules, while the processing sequence on the decoding side can be implemented by interaction of these modules.

[71] In the known method, if the current block is bidirectional, then there is no rational solution to the problem of how to correct the first and second original motion vectors after obtaining the first and second original motion vectors of the current block. In an embodiment of the application, if the current block is a bidirectional block, then it is first judged whether the characteristic information of the current block meets a specific condition; if so, determine the first reference block corresponding to the current block based on the first original motion vector, determine the second reference block based on the second original motion vector of the current block, and adjust the first and second original motion vectors according to the first pixel value of the first reference block and the second value pixels of the second reference block to obtain the first and second target motion vectors.

[72] Thus, the first and second target motion vectors are obtained by correcting the first and second source motion vectors, which makes it possible to encode or decode the current block based on the first and second target motion vectors, improve encoding performance, and improve encoding efficiency.

[73] Below, the method of encoding and decoding according to the present application is described in detail using several particular embodiments as an example.

[74] Embodiment 1: FIG. 2 is a flowchart of an encoding and decoding method according to an embodiment of the present application, the method is applicable to a decoding side or an encoding side, and the method may include the following steps:

[75] In step 201, if the characteristic information of the current block meets the specific condition, the first reference block corresponding to the current block is determined based on the first original motion vector of the current block, and the second reference block corresponding to the current block is determined based on the second original motion vector of the current block. block. The characteristic information includes, but is not limited to, at least one of the following: a motion information prediction mode corresponding to the current block; a motion information attribute corresponding to the current block; information about the size of the current block.

[76] In one example, when the characteristic information is the motion information prediction mode corresponding to the current block, matching the characteristic information of the current block to a particular condition may include, but is not limited to: if the motion information prediction mode corresponding to the current block is standard merge mode, determining that the motion information prediction mode corresponding to the current block meets a specific condition; or, if the motion information prediction mode corresponding to the current block is a merge mode to generate a new prediction value by combining the inter prediction value and the intra prediction value, it is determined that the motion information prediction mode corresponding to the current block meets the special condition.

[77] In one example, when the characteristic information is a motion information attribute corresponding to the current block, matching the characteristic information of the current block to a specific condition may include, but is not limited to: if the motion information attribute corresponding to the current block is, that the motion information of the current block includes motion information in two different directions, determining that the motion information attribute corresponding to the current block meets a specific condition; or, if the motion information attribute corresponding to the current block is that the motion information of the current block includes motion information in two different directions, and the distances between the current frame and each of the two reference frames corresponding to the motion information in two different directions are the same, determine that the motion information attribute corresponding to the current block meets the particular condition; or, if the motion information attribute corresponding to the current block is that for the current block, the motion information of surrounding blocks is reused, determining that the motion information attribute corresponding to the current block meets a special condition; or, if the motion information attribute corresponding to the current block is that the motion information of each sub-block of the current block is the same, it is determined that the motion information attribute corresponding to the current block meets the special condition.

[78] For example, that the motion information of the current block includes motion information in two different directions means: the motion information of the current block in two different directions refers to the first reference frame and the second reference frame, respectively, with the first reference the frame is located before the current frame in which the current block is located, and the second reference frame is located behind the current frame in which the current block is located.

[79] In one example, when the characteristic information is current block size information, the current block size information may include a width value and a height value of the current block, wherein the characteristic information of the current block matches a particular condition may include, in addition to other things: if the width value of the current block is within the first interval [first threshold, second threshold], determining that the size information of the current block meets a specific condition; or, if the height value of the current block is within the second interval [third threshold, fourth threshold], determining that the size information of the current block meets a special condition; or, if the area obtained based on the width value and the height value of the current block is within the third interval [fifth threshold, sixth threshold], it is determined that the size information of the current block meets a specific condition; or, if the width value is within the first interval [first threshold, second threshold], the height value is within the second interval [third threshold, fourth threshold], and the area is within the third interval [fifth threshold, sixth threshold], determine, that the information about the size of the current block meets a special condition.

[80] For example, the first threshold may be less than the second threshold, wherein neither the first threshold nor the second threshold is limited, for example, the first threshold may be 4 and the second threshold may be 128. The third threshold may be less than the fourth threshold, wherein neither the third threshold nor the fourth threshold is limited, for example, the third threshold may be 8, and the fourth threshold may be 128. The fifth threshold may be less than the sixth threshold, and neither the fifth nor the sixth threshold is limited, for example, the the threshold may be 64 and the sixth threshold may be 128*128.

[81] In one example, determining a first reference block corresponding to the current block based on the first original motion vector of the current block and determining a second reference block corresponding to the current block based on the second original motion vector of the current block may include , inter alia, the actions of: determining the first reference block corresponding to the current block from the first reference frame based on the first source motion vector of the current block; and determining a second reference block corresponding to the current block from the second reference frame based on the second source motion vector of the current block. The first pixel value of each pixel point in the first reference block is obtained by interpolating the pixel values of the adjacent pixel points in the first reference block, or obtained by copying the pixel values of the adjacent pixel points in the first reference block; the second pixel value of each pixel point in the second reference block is obtained by interpolating the pixel values of adjacent pixel points in the second reference block, or obtained by copying the pixel values of adjacent pixel points in the second reference block. The size of the first reference block is equal to the size of the second reference block, while the width of the first reference block is determined by the width of the current block and the search range, and the height of the first reference block is determined by the height of the current block and the search range.

[82] In step 202, the first and second source motion vectors are adjusted according to the first pixel value of the first reference block and the second pixel value of the second reference block to obtain a first target motion vector corresponding to the first source motion vector and a second target motion vector corresponding to the second source vector. movement.

[83] In one example, if the current block contains at least one subblock, then for each subblock of the current block, the first and second source motion vectors are adjusted according to the first and second pixel values to obtain the first and second target motion vectors of the subblock.

[84] In one example, the step of adjusting the first and second source motion vectors according to the first and second pixel values to obtain the first and second target motion vectors of the subblock may include determining the first and second integer pixel sub-block motion vector correction values; and/or first and second sub-pixel sub-block motion vector correction values based on the first and second pixel values. Further, the first source motion vector may be corrected according to the first integer-pixel motion vector correction value and/or the first sub-pixel motion vector correction value to obtain the first sub-block target motion vector. Meanwhile, the second source motion vector can be corrected according to the second integer-pixel motion vector correction value and/or the second sub-pixel motion vector correction value to obtain the second sub-block target motion vector.

[85] In one example, determining the first and second integer sub-block motion vector correction values, the first and second sub-pixel sub-block motion vector correction values based on the first and second pixel values, may include: determining the first source motion vector or the second source motion vector as the center motion vector; determining an edge motion vector corresponding to the central motion vector; obtaining a first cost value corresponding to the center motion vector and a second cost value corresponding to the edge motion vector based on the first and second pixel values; next, selecting a motion vector from the center motion vector and the edge motion vector as the optimal motion vector based on the first cost value and the second cost value, and judging whether the end condition is met; if not, determining the optimal motion vector as the center motion vector, and re-determining an edge motion vector corresponding to the center motion vector; if so, determining first and second integer sub-unit motion vector correction values based on the optimal motion vector; and determining first and second sub-pixel sub-block motion vector correction values based on the optimal motion vector.

[86] In one example, in a step in which the edge motion vector corresponding to the center motion vector is determined: the center motion vector (x, y) is shifted in different directions by S, and the edge motion vector (x, y+S) is sequentially obtained, the edge vector (x, y-S) of movement, edge vector (x+ S, y) of movement, edge vector (x-S, y) of movement and edge vector (x+ right, y+ down) of movement in different directions, or shift the center vector (x, y) move in different directions on S and sequentially obtain the edge vector (x, y-S) of the movement, the edge vector (x, y+S) of the movement, the edge vector (x-S, y) of the movement, the edge vector of (x+S, y) of the movement and the edge vector (x+ right, y+ down) of movement in different directions. The default edge vector value of (x + right, y + down) movement is (x-S, y-S); if the cost value of the edge vector (x+S, y) of the movement is less than the cost value of the edge vector (x-S, y) of the movement, then the value of "right" is S; if the cost value of the edge motion vector (x, y+S) is less than that of the edge motion vector (x, y-S), then the "down" value is S; or, if the cost value of the edge vector (x+S, y) of the motion is less than or equal to the value of the cost of the edge vector (x-S, y) of the motion, then the value of "right" is S; if the edge motion vector cost value (x, y + S) is less than or equal to the edge motion vector cost value (x, y-S) then the down value is S.

[87] In one example, the step of obtaining a first cost value corresponding to a center motion vector and a second cost value corresponding to an edge motion vector based on the first and second pixel values may include, among other things, the steps of that: obtaining a first cost value corresponding to a center motion vector and a second cost value corresponding to an edge motion vector based on the first and second pixel values without downsampling; alternatively, a downsampling operation may be performed on the first and second pixel values; and obtaining a first cost value corresponding to a center motion vector and a second cost value corresponding to an edge motion vector based on the first and second downsampled pixel values; alternatively, you can perform a shift and downsampling operation on the first pixel value, and perform a shift and downsampling operation on the second pixel value; and obtaining a first cost value corresponding to the center motion vector and a second cost value corresponding to the edge motion vector based on the first and second pixel values after shifting and downsampling.

[88] In one example, determining the first and second integer sub-block motion vector correction values, the first and second sub-pixel sub-block motion vector correction values based on the first and second pixel values, may include the steps of: taking the first source motion vector or the second source motion vector as the center, selecting some or all of the motion vectors from the surrounding motion vectors, and taking the selected motion vectors as candidate motion vectors; obtaining a third cost value corresponding to the first source motion vector or the second source motion vector and a fourth cost value corresponding to each of the potential motion vectors based on the first and second pixel values; then selecting a motion vector from the first original motion vector or the second original motion vector and the potential motion vector as the optimal motion vector based on the third cost value and the fourth cost value; determining first and second integer subblock motion vector correction values based on the optimal motion vector; and determining first and second sub-pixel sub-block motion vector correction values based on the optimal motion vector.

[89] The step of determining the first and second integer pixel motion vector correction values of the sub-unit based on the optimal motion vector may include, among other things, the steps of: determining the first integer pixel motion vector correction value of the sub-block based on the optimal the motion vector and the first original motion vector, and determining a second integer pixel motion vector correction value of the subblock according to the first integer pixel motion vector correction value.

[90] The step of determining the first and second sub-pixel motion vector correction values of the sub-block based on the optimal motion vector may include: determining the first sub-pixel motion vector correction value of the sub-block based on the cost value corresponding to the optimal motion vector and a cost value corresponding to an edge motion vector corresponding to the optimal motion vector, and determining a second sub-pixel motion vector correction value of the subblock based on the first sub-pixel motion vector correction value.

[91] At step 203, the current block is encoded or decoded based on the first and second target motion vectors. For example, when it comes to the encoding side, the encoding side may encode the current block based on the first and second target motion vectors; when it comes to the decoding side, the decoding side may decode the current block based on the first and second target motion vectors.

[92] In one example, for each subblock of the current block, a third reference block corresponding to the subblock is determined based on the first target motion vector of the subblock, and a fourth reference block corresponding to the subblock is determined based on the second target motion vector of the subblock; the third pixel value of the third reference block and the fourth pixel value of the fourth reference block are weighted to obtain a sub-block prediction value; and determining a prediction value of the current block based on the prediction value of each sub-block.

[93] In one example, determining a third reference block corresponding to a subblock based on the first target motion vector of the subblock and determining a fourth reference block corresponding to the subblock based on the second target motion vector of the subblock may include: , what:

[94] Embodiment 1: Determine the fifth reference block corresponding to the sub-block from the first reference frame based on the first target motion vector of the sub-block, and interpolate pixel values in the fifth reference block to obtain the third reference block. The sixth reference block corresponding to the subblock is determined from the second reference frame based on the second target motion vector of the subblock, and the pixel values in the sixth reference block are interpolated to obtain the fourth reference block.

[95] Embodiment 2: Determine the seventh reference block corresponding to the sub-block from the first reference frame based on the first target motion vector of the sub-block, build the eighth reference block based on the pixel values in the seventh reference block, and interpolate the pixel values in the eighth reference block to obtain the third support block; and determining a ninth reference block corresponding to the subblock from the second reference frame based on the second target motion vector of the subblock, constructing a tenth reference block based on the pixel values in the ninth reference block, and interpolating the pixel values in the tenth reference block to obtain the fourth reference block.

[96] In one example, after encoding or decoding the current block based on the first and second target motion vectors, the first and second target motion vectors for the current block may be stored. Thus, the first and second target motion vectors are used for contour filtering of the current frame, the first and second target motion vectors are used for anchoring in the time domain of subsequent frames, and/or the first and second target motion vectors are used for anchoring in the spatial domain of the current frame.

[97] It follows from the above disclosed technical solutions that in the embodiment of the present application, the first and second target motion vectors of the current block can be determined based on the first and second source motion vectors, and the current block can be encoded or decoded based on the first and second target motion vectors instead of encoding or decoding it directly based on the first and second original motion vectors, to solve problems such as poor prediction quality, prediction error, and the like, improve encoding performance, and improve encoding efficiency.

[98] Embodiment 2: FIG. 3 is a flowchart of the encoding and decoding method of an embodiment of the present application, wherein the encoding and decoding method is applicable to the encoding side, and the encoding and decoding method may include the following steps:

[99] In step 301, the encoding side judges whether the characteristic information of the current block meets a specific condition. If the characteristic information of the current block meets the specific condition, step 302 is executed; and if the characteristic information of the current block does not meet the specific condition, it is not necessary to apply the motion vector correction method of the present application, and the processing method is not limited.

[100] In one example, the encoding side may judge whether the characteristic information of the current block meets a particular condition based on the encoded information. If the characteristic information of the current block meets the special condition, it indicates that the motion information of the current block is not accurate enough, and therefore the motion vector refinement mode is turned on (enabled), and step 302 is performed.

[101] If the characteristic information of the current block does not meet a special condition, this indicates that the motion information of the current block is sufficiently accurate, so the motion vector refinement mode is not allowed, and the motion vector refinement mode of the present application is not necessary.

[102] In one example, the characteristic information of the current block includes, among other things, at least one of the following: a motion information prediction mode corresponding to the current block; a motion information attribute corresponding to the current block; information about the size of the current block.

[103] Embodiments of the method for assessing whether the characteristic information of the current block meets a particular condition:

[104] In step 302, the encoding side determines the first reference block corresponding to the current block from the first reference frame based on the first source motion vector of the current block; and determines a second reference block corresponding to the current block from the second reference frame based on the second source motion vector of the current block. The pixel value of each pixel point in the first reference block is referred to as the "first pixel value", and the pixel value of each pixel point in the second reference block is referred to as the "second pixel value" to differentiate between them.

[105] In one example, if the current block is a block for which bidirectional prediction is applied, then the motion information of the current block is bidirectional with two reference frames and two original motion vectors. For example, the encoding side may obtain bi-directional motion information in a known manner, which is not limited. The bidirectional motion information includes a first reference frame and a first source motion vector, a second reference frame and a second source motion vector.

[106] The encoding side may determine the first reference block corresponding to the current block from the first reference frame based on the first original motion vector, wherein the pixel value of each pixel point in the first reference block may be referred to as the "first pixel value".

[107] The encoding side may determine a second reference block corresponding to the current block from the second reference frame based on the second original motion vector, wherein the pixel value of each pixel point in the second reference block may be referred to as a “second pixel value”.

[108] In one example, the distance between the current frame in which the current block is located and the first reference frame and the distance between the second reference frame and the current frame in which the current block is located may be the same. For example, the first key frame is the 1st frame, the current frame is the 5th frame, and the second key frame is the 9th frame. Of course, the above is just one example, and the two distances may be different.

[109] The first and second source motion vectors are mirror symmetrical to each other, for example, the first source motion vector has coordinates (4, 4), and the second source motion vector has coordinates (-4, -4); the first source motion vector has coordinates (2.5, 3.5) and the second source motion vector has coordinates (-2.5, -3.5). Of course, the above is only one of the examples, and there may not be mirror symmetry between the first and second initial motion vectors.

[110] Embodiments of the method for determining the first and second reference blocks:

[111] In step 303, for each subblock of the current block: the encoding side adjusts the first source motion vector according to the first pixel value of the first reference block and the second pixel value of the second reference block to obtain the first target motion vector of the subblock; and the encoding side corrects the second source motion vector according to the first pixel value of the first reference block and the second pixel value of the second reference block to obtain a second sub-block target motion vector.

[112] In one example, if the motion vector refinement mode is enabled (enabled) for the current block, the encoding side may fine-tune the first and second source motion vectors using a local search method on the first and second pixel values to obtain more correct first target vector motion and the second target motion vector, and then generate a prediction value with less distortion based on them.

[113] In one example, the current block may include at least one subblock, and if it includes only one subblock, then that subblock is the current block. This sub-block corresponds to the first and second source motion vectors, and after their correction, the sub-block corresponds to the first and second target motion vectors.

[114] If the current block includes sub-block A and sub-block B, then sub-block A corresponds to the first source motion vector A1 and the second source motion vector A2, and after their correction, sub-block A corresponds to the first target motion vector A3 and the second target motion vector A4. The sub-block B corresponds to the first source motion vector B1 and the second source motion vector B2, and after their correction, the sub-block B corresponds to the first target motion vector B3 and the second target motion vector B4.

[115] The first original motion vector A1 corresponding to sub-block A and the first original motion vector B1 corresponding to sub-block B may be the same and be the first original motion vectors of the current block; the second source motion vector A2 corresponding to sub-block A and the second source motion vector B2 corresponding to sub-block B may be the same and be the second source motion vector of the current block.

[116] Since the first source motion vector of each subblock is adjusted separately, the first target motion vector A3 corresponding to subblock A and the first target motion vector B3 corresponding to subblock B may be the same or different vectors. Since the second source motion vector of each subblock is adjusted separately, the second target motion vector A4 corresponding to subblock A and the second target motion vector B4 corresponding to subblock B may be the same or different vectors.

[117] Embodiments of the method for correcting the original motion vectors, which will not be detailed in this paragraph, see below.

[118] In step 304, the encoding side performs motion compensation for each subblock of the current block based on the first and second target motion vectors. For example, if the current block includes sub-block A and sub-block B, the first and second target motion vectors of sub-block A may serve to perform motion compensation for sub-block A, and the first and second target motion vectors of sub-block B may serve to perform motion compensation for sub-block B. .

[119] Embodiments of the motion compensation method, which will not be detailed in this paragraph, see below.

[120] In step 305, the encoding side stores the first and second target motion vectors of each subblock of the current block, with the first and second target motion vectors being used as references for encoding the subsequent block.

[121] For implementations of support vector coding, see the following embodiments, which will not be detailed in this paragraph.

[122] Embodiment 3: FIG. 4 is a flowchart of an encoding and decoding method according to an embodiment of the present application, wherein the encoding and decoding method is applicable to the decoding side, and the encoding and decoding method may include the following steps:

[123] In step 401, the decoding side judges whether the characteristic information of the current block meets a specific condition. If the characteristic information of the current block meets the special condition, then step 402 is executed, and if the characteristic information of the current block does not meet the special condition, then there is no need to apply the motion vector correction method of the present application, and the processing method is not limited.

[124] In one example, the decoding side is configured to receive the bitstream sent by the encoding side and determine whether the characteristic information of the current block meets a particular condition based on the decoded information. If the characteristic information of the current block meets the special condition, this indicates that the motion information of the current block is not accurate enough, and therefore the motion vector refinement mode is turned on and step 402 is performed. If the characteristic information of the current block does not meet the special condition, this indicates that the motion information of the current block is accurate enough, therefore, the motion vector refinement mode is not turned on, and the motion vector refinement mode is not necessary in the present application.

[125] In step 402, the decoding side determines the first reference block corresponding to the current block from the first reference frame based on the first source motion vector of the current block; and determines a second reference block corresponding to the current block from the second reference frame based on the second source motion vector of the current block. The pixel value of each pixel point in the first reference block is referred to as the "first pixel value", and the pixel value of each pixel point in the second reference block is referred to as the "second pixel value" to differentiate between them.

[126] In step 403, for each subblock of the current block: the decoding side adjusts the first source motion vector according to the first pixel value of the first reference block and the second pixel value of the second reference block to obtain the first target motion vector of the subblock; and the decoding side corrects the second source motion vector according to the first pixel value of the first reference block and the second pixel value of the second reference block to obtain a second sub-block target motion vector.

[127] In step 404, the decoding side performs motion compensation for each subblock of the current block based on the first and second target motion vectors of the subblock.

[128] In step 405, the decoding side stores the first and second target motion vectors of each sub-block of the current block, with the first and second target motion vectors used as references for encoding the subsequent block.

[129] Embodiment 4: In step 301 and step 401, it is judged whether the characteristic information of the current block meets a specific condition. In this embodiment, if the characteristic information meets all of the following conditions, it is determined that the characteristic information of the current block meets the specific condition, and the motion vector refinement mode is turned on. Otherwise, the motion vector refinement mode is not turned on.

[130] The motion information attribute corresponding to the current block includes the following: directly multiplexing in the current block the motion information of the surrounding blocks without encoding the motion information difference. The motion information attribute corresponding to the current block includes the following: the motion information of each sub-block of the current block is the same, i. e. without applying the sub-block motion information prediction mode. The motion information attribute corresponding to the current block includes the following: the motion information of the current block includes motion information in two different directions. The information about the size of the current block includes: the size of the current block is within a certain range, as will be disclosed in the following embodiments.

[131] Embodiment 5: In step 301 and step 401, it is judged whether the characteristic information of the current block meets a particular condition. In this embodiment, if the characteristic information meets all of the following conditions, it is determined that the characteristic information of the current block meets the specific condition, and the motion vector refinement mode is turned on. Otherwise, the motion vector refinement mode is not turned on.

[132] The motion information prediction mode corresponding to the current block includes a regular merge mode or a CIIP mode (namely, a merge mode for generating a new prediction value by combining an inter prediction value and an intra prediction value); alternatively, the motion information prediction mode corresponding to the current block does not include an MMVD mode (i.e., a merge mode for encoding motion information difference values), an SB merge mode (i.e., a merge mode for applying motion information subblock motion) or TPM mode (i.e. merge mode for triangle prediction).

[133] The motion information attribute corresponding to the current block includes the following: the motion information of the current block includes motion information in two different directions, wherein the distances between the current frame and each of the two reference frames corresponding to the motion information in two different directions are the same. Information about the size of the current block includes: the size of the current block is within a certain range.

[134] It should be noted that the merge mode includes, but is not limited to, the following types of merge modes: standard merge mode (i.e., standard merge mode); merge mode for triangle prediction (TPM mode); a merge mode for encoding motion information difference values (MMVD mode); a merging mode for applying sub-unit motion information (SB merging mode); and a merge mode for generating a new prediction value by combining the inter prediction value and the intra prediction value (CIIP mode).

[135] In one example, the standard merge mode includes a standard merge mode in which residues are not encoded (ie, skip mode) and a standard merge mode in which residues are encoded. The MMVD mode includes an MMVD mode in which the remainder is not encoded.

[136] Embodiment 6: In step 301 and step 401, it is judged whether the characteristic information of the current block meets a particular condition. In this embodiment, if the characteristic information meets all of the following conditions, it is determined that the characteristic information of the current block meets the specific condition, and the motion vector refinement mode is turned on. Otherwise, the motion vector refinement mode is not turned on.

[137] The motion information prediction mode corresponding to the current block includes a standard merge mode (i.e., a standard merge mode), a CIIP mode (namely, a merge mode for generating a new prediction value by combining an inter prediction value and an intra prediction value ) or TPM mode (i.e. merge mode for triangle prediction); alternatively, the motion information prediction mode corresponding to the current block does not include an MMVD mode (i.e., a merge mode for encoding motion information difference values) or an SB merge mode (i.e., a merge mode for applying motion information). subblock movement).

[138] The motion information attribute corresponding to the current block includes the following: the motion information of the current block includes motion information in two different directions, wherein the distances between the current frame and each of the two reference frames corresponding to the motion information in two different directions are the same. Information about the size of the current block includes: the size of the current block is within a certain range.

[139] Embodiment 7: In step 301 and step 401, it is judged whether the characteristic information of the current block meets a particular condition. In this embodiment, if the characteristic information meets all of the following conditions, it is determined that the characteristic information of the current block meets the specific condition, and the motion vector refinement mode is turned on. Otherwise, the motion vector refinement mode is not turned on.

[140] The motion information prediction mode corresponding to the current block includes a standard merge mode (i.e., a standard merge mode), a CIIP mode (namely, a merge mode for generating a new prediction value by combining an inter prediction value and an intra prediction value ), TPM mode (i.e., merge mode for triangle prediction), or MMVD mode (i.e., merge mode for encoding motion information difference values); alternatively, the motion information prediction mode corresponding to the current block does not include the SB merge mode (ie, the merge mode for applying sub-block motion information).

[141] The motion information attribute corresponding to the current block includes the following: the motion information of the current block includes motion information in two different directions, wherein the distances between the current frame and each of the two reference frames corresponding to the motion information in two different directions are the same. Information about the size of the current block includes: the size of the current block is within a certain range.

[142] Embodiment 8: In step 301 and step 401, it is judged whether the characteristic information of the current block meets a particular condition. In this embodiment, if the characteristic information meets all of the following conditions, it is determined that the characteristic information of the current block meets the specific condition, and the motion vector refinement mode is turned on. Otherwise, the motion vector refinement mode is not turned on.

[143] The motion information prediction mode corresponding to the current block includes a standard merge mode or a CIIP mode; the motion information prediction mode corresponding to the current block does not include the MMVD mode, the SB merge mode, or the TPM mode; alternatively, the motion information prediction mode corresponding to the current block includes a standard merge mode, a CIIP mode, or a TPM mode; the motion information prediction mode corresponding to the current block does not include the MMVD mode or the SB merge mode; alternatively, the motion information prediction mode corresponding to the current block includes a standard merge mode, a CIIP mode, or a TPM mode, or an MMVD mode; the motion information prediction mode corresponding to the current block does not include the SB merge mode.

[144] The motion information attribute corresponding to the current block includes the following: the motion information of the current block includes motion information in two different directions, and there is no limitation that the distances between the current frame and each of two reference frames corresponding to motion information in two different directions are the same, i.e. this limiting condition has been removed. Information about the size of the current block includes: the size of the current block is within a certain range.

[145] Embodiment 9: Regarding "the size of the current block is within a certain range" in the above-disclosed Embodiments 4 to 8: if the size of the current block meets the following condition, it is determined that the size of the current block is within a certain range.

[146] The width value of the current block is within the range of [first threshold, second threshold], wherein the first threshold may be 4 and the second threshold may be 128. The height of the current block is within the range of [third threshold, fourth threshold], wherein the third the threshold may be 8 and the fourth threshold may be 128. For example, each of the first threshold, second threshold, third threshold, and fourth threshold may be ²ⁿ , where n is an integer greater than or equal to 1. The area of the current block is within the range [fifth threshold , sixth threshold] where the fifth threshold could be 64 and the sixth threshold could be 128*128= 16384. For example, both the fifth threshold and the sixth threshold could be 4 ^m , where m is an integer greater than or equal to 1. In the above example, the expression [a, b] can represent a value no less than a and no greater than b.

[147] Embodiment 10: At step 302 and step 402, the first reference block corresponding to the current block is determined from the first reference frame based on the first source motion vector, wherein the pixel value of each pixel point in the first reference block is referred to as the "first pixel value" ; determining a second reference block corresponding to the current block from the second reference frame based on the second source motion vector, wherein the pixel value of each pixel point in the second reference block is referred to as the "second pixel value", as discussed below.

[148] The first pixel value of each pixel point in the first reference block is obtained by interpolating the pixel values of the adjacent pixel points in the first reference block, or obtained by copying the pixel values of the adjacent pixel points in the first reference block; the second pixel value of each pixel point in the second reference block is obtained by interpolating the pixel values of adjacent pixel points in the second reference block, or obtained by copying the pixel values of adjacent pixel points in the second reference block. The size of the first reference block is equal to the size of the second reference block, the width value of the first/second reference block is determined from the width value of the current block and the search range, and the height value of the first/second reference block is determined from the height value of the current block and the search range.

[149] In Fig. 5, if the current block has a width W and a height H, then the first source motion vector is designated MV0 and the second source motion vector is designated MV1. An integer pixel block of area (W+FS-1)*(H+FS-1) is obtained based on the corresponding position of the first source motion vector MV0 in the first reference frame, the integer pixel block being referred to as integer pixel block A. An integer pixel block of area (W+FS -1)*(H+FS-1) is obtained based on the corresponding position of the second source motion vector MV1 in the second reference frame, the integer pixel block being referred to as integer pixel block B.

[150] In one example, based on an integer block A of area (W+FS-1)*(H+FS-1), an original block of reference pixels of size (W +2*SR)*(H +2*SR) can be obtained by bilinear interpolation, wherein the original block of reference pixels may be referred to as "first reference block". On the basis of an integer block B of area (W+FS-1)*(H+FS-1), you can get the original block of reference pixels of size (W +2*SR)*(H +2*SR) by bilinear interpolation, while the original the block of reference pixels can be referred to as the "second reference block".

[151] In another example, based on an integer block A of area (W+FS-1)*(H+FS-1), an original block of reference pixels of size (W +2*SR)*(H +2*SR) can be obtained by direct copying (without interpolation), while the original block of reference pixels can be referred to as the "first reference block". Based on the whole-pixel block B of area (W+FS-1)*(H+FS-1), you can get the original block of reference pixels of size (W +2*SR)*(H +2*SR) by direct copying, while the original block of reference pixels can be referred to as the "second reference block".

[152] For example, for the luminance component only (since in the subsequent search process, for simplicity, the cost value will be calculated only for the luminance component), you can get the original blocks of reference pixels of size (W + 2 * SR) * (H + 2 * SR) based on integer pixel blocks (for example, integer pixel block A and integer pixel block B) of area (W+FS-1)*(H+FS-1), i.e. the first reference block (eg Pred_Inter0) and the second reference block (eg Pred_Inter1).

[153] In one example, FS may be the number of interpolation filter taps, such as FS may be 8, and so on.

[154] Obtaining the first/second reference blocks by bilinear interpolation is to obtain the pixel value of each pixel point in the first/second reference block by interpolating the pixel values of adjacent pixel points in the first/second reference block, but is not limited to this. Obtaining the first/second reference blocks by copying is that the pixel value of each pixel point in the first/second reference block is obtained by copying the pixel values of adjacent pixel points in the first/second reference block, but is not limited to this.

[155] In the above embodiment, the first reference block has an area of (W+2*SR)*(H+2*SR), the second reference block has an area of (W+2*SR)*(H+2*SR), then there is a width value of the first/second reference block is W+2*SR, and a height value of the first/second reference block is H+2*SR. W is the width of the current block, H is the height of the current block, SR is the search range, i.e. the number of iterations in the following embodiments; thus, SR is the maximum interpolation over the horizontal/vertical components of the target motion vector and the source motion vector, eg SR could be 2, and so on.

[156] The first and second reference blocks are used to correct the motion vector in the subsequent process.

[157] Embodiment 11: in step 303 and step 403, for each sub-block of the current block, adjust the first source motion vector according to the first pixel value of the first reference block and the second pixel value of the second reference block to obtain the first target motion vector of the sub-block; while correcting the second source motion vector according to the first pixel value of the first reference block and the second pixel value of the second reference block to obtain the second target motion vector of the subblock.

[158] As an example, the original motion vector correction process is disclosed as a procedure for processing one sub-block (eg, each sub-block of size dx*dy of the current block, for example, a sub-block of 16*16 or smaller block).

[159] In step a1, the first original motion vector or the second original motion vector is determined as the center motion vector.

[160] For example, if the first source motion vector has coordinates (4, 4) and the second source motion vector has coordinates (-4, -4), then the first source motion vector (4, 4) or the second source motion vector (- 4, -4) can be defined as the central motion vector. For convenience of disclosure, for example, the first original motion vector (4, 4) is determined as the central motion vector, while the procedure for determining the second initial motion vector (-4, -4) as the central motion vector is similar.

[161] In step a2, an edge motion vector corresponding to the center motion vector is determined.

[162] For example, the center motion vector (x, y) is shifted in different directions by S, and edge motion vector (x, y+S), edge motion vector (x, y-S), edge vector (x+ S, y) movement, an edge vector (x-S, y) of movement, and an edge vector (x+ right, y+ down) of movement in different directions, where the value "right" can be S or -S, and the value "down" can be S or -S, with the definition of the values "right" and "down" are disclosed in the following embodiments.

[163] In Fig. 6, for example, if the central motion vector (x, y) is taken as the center, that is, the central motion vector has coordinates (0, 0), S is 1, and the values "right" and "down" are 1, then the number of edge motion vectors corresponding to the central motion vector (0, 0) includes: edge vector (0, 1) motion, edge vector (0, -1) motion, edge vector (1, 0) motion, edge vector (-1, 0) of motion and edge vector (1, 1) of motion.

[164] In step a3, a first cost value corresponding to the center motion vector and a second cost value corresponding to each edge motion vector are obtained based on the first pixel value of the first reference block and the second pixel value of the second reference block.

[165] For example, the reference subblock A1 corresponding to the central motion vector (0, 0) is obtained by copying from the first reference block, wherein the reference subblock A1 is the reference subblock of the central motion vector (0, 0) in the first reference block. The size of the reference sub-block A1 is the size of the current block, with the reference sub-block A1 being the closest to the center block W*H in the first reference block.

[166] Reference subblock B1 corresponding to the center motion vector (0, 0) is obtained by copying from the second reference block, wherein reference subblock B1 is the reference subblock of the central motion vector (0, 0) in the second reference block. The B1 reference subblock size is the size of the current block, with the B1 reference subblock being the closest to the center W*H block in the second reference block.

[167] Next, a cost value 1 corresponding to the center motion vector (0, 0) is obtained based on the first pixel value of the reference subblock A1 and the second pixel value of the reference subblock B1; the value value determination method is disclosed in the following embodiments.

[168] Reference subblock A2, (0, 1) motion is obtained by copying from the first reference block, wherein reference subblock A2 is the reference subblock of edge vector (0, 1) motion in the first reference block. The size of the reference sub-block A2 is the size of the current block, with the reference sub-block A2 being the W*H block closest to the center in the first reference block.

[169] Reference subblock B2 corresponding to motion vector (0, -1) symmetric edge vector (0, 1) of motion is obtained by copying from the second reference block, wherein reference subblock B2 is a reference subblock of symmetric vector (0, - 1) movement in the second support block. The size of the reference subblock B2, which is the W*H block in the second reference block, is the size of the current block, at.

[170] Next, a cost value of 2 corresponding to an edge motion vector (0, 1) is obtained based on the first pixel value of the reference subblock A2 and the second pixel value of the reference subblock B2, and the cost value determination method is disclosed in the following embodiments.

[171] According to this method, it is possible to determine a cost value 2 corresponding to the edge motion vector (0, 1), a cost value 3 corresponding to the motion edge vector (0, −1), a cost value 4 corresponding to the motion edge vector (1, 0). , a cost value 5 corresponding to the edge vector (-1, 0) of the motion, and a cost value 6 corresponding to the edge vector (1, 1) of the motion.

[172] In step a4: a motion vector is selected from the center motion vector and the edge motion vector as the optimal motion vector based on the first cost value and the second cost value. For example, the motion vector with the lowest cost value can serve as the optimal motion vector. For example, if the cost value 2 corresponding to the edge motion vector (0, 1) is the minimum, the edge motion vector (0, 1) corresponding to the cost value 2 can serve as the optimal motion vector.

[173] At step a5: evaluate compliance with the end condition; in case of non-compliance with the end condition, the optimal motion vector is determined as the central motion vector, return to step a2 and its execution; if the end condition is met, step a6 is performed.

[174] In one example, when a threshold of iterations/search range is reached, the end condition is met; if the threshold for the number of iterations / search range is not reached, there is a mismatch in the end condition. For example, if the value of the search range (SR, English search range) is 2, i.e. the threshold value is 2, the allowed number of iterations is two. If the number of iterations / search range value reaches 2, i.e. steps a2 - a4 will be executed twice, then the end condition is met; otherwise, there is a discrepancy with the final condition.

[175] In an additional example, after selecting a motion vector from the center motion vector and edge motion vectors as the optimal motion vector, if the center motion vector is selected as the optimal motion vector, then the end condition will be met.

[176] In step a6, a first integer pixel motion vector correction value (which is used to correct the first original motion vector) and a second integer pixel motion vector correction value (which is used to correct the second original motion vector) are determined based on the optimal motion vector.

[177] In one example, the first integer pixel motion vector correction value may be determined based on the optimal motion vector and the first original motion vector, and the second integer pixel motion vector correction value may be determined according to the first integer pixel motion vector correction value, wherein the second integer pixel motion vector correction value is symmetrical the first integer value of the motion vector correction.

[178] For example, if during the first iteration the optimal motion vector is the edge motion vector (0, 1), then the second iteration is performed with the edge vector (0, 1) as the center, while during the second iteration the optimal motion vector is the edge vector (0, 1) of the motion; if the iterative process ends there, then the first integer pixel motion vector correction value is (0, 2), that is, the sum of the (0, 1) motion edge vector and the (0, 1) motion edge vector.

[179] In view of the above, if the first initial motion vector has coordinates (4, 4), then the optimal motion vector is the edge vector (0, 1) of motion during the first iteration, i.e. the optimal motion vector may correspond to the optimal motion vector (4, 5). When performing the process of the second iteration, the edge vector (0, 1) of the motion is taken as the center, and the optimal motion vector is the edge vector (0, 1) of the motion in the second iteration, that is, the optimal motion vector can correspond to the optimal vector (4, 6 ) movement.

[180] Thus, the first integer pixel motion vector correction value is determined based on the optimal motion vector (4, 6) and the first original motion vector (4, 4), wherein the first integer pixel motion vector correction value is the difference of the optimal motion vector (4, 6) movement and the first initial vector (4, 4) of movement, i.e. the first integer pixel value of the motion vector correction is (0, 2).

[181] Next, a second integer pixel motion vector correction value is determined according to the first integer pixel motion vector correction value (0, 2), which may be (0, -2), i. be a value symmetric to (0, 2).

[182] In step a7, a first sub-pixel motion vector correction value (which is used to correct the first original motion vector) and a second sub-pixel motion vector correction value (which is used to correct the second original motion vector) are determined based on the optimal motion vector.

[183] In one example, the first sub-pixel motion vector correction value can be determined based on the cost value corresponding to the optimal motion vector and the cost value corresponding to the edge motion vector corresponding to the optimal motion vector, and then to determine the second sub-pixel motion vector correction value based on the first subpixel motion vector correction value.

[184] For example, x ₀ =N*(E(-1.0) - E(1.0)) / (E(-1.0) + E(1.0) - 2* E(0.0 )), y ₀ =N*(E(0,-1) - E(0.1)) / (E(0,-1) + E(0.1) -2* E(0.0)) , and if the motion vector pixel precision is 1/2, 1/4, 1/8, and 1/16, then N=1, 2, 4, and 8, respectively. Next, (x ₀ , y ₀ ) is assigned to deltaMv, SPMV = deltaMv/2N, and if the pixel precision of the current motion vector is 1/16, then SPMV is (x ₀ /16, y ₀ /16).

[185] In the above equation, SPMV is the first sub-pixel motion vector correction value, and N is related to the pixel precision of the motion vector, for example, if the pixel precision of the motion vector is 1/2, then N is 1; if the pixel precision of the motion vector is 1/4, then N is 2; if the pixel precision of the motion vector is 1/8, then N is 4; if the pixel precision of the motion vector is 1/16, then N is 8.

[186] In the above equation, E(0,0) is the cost value of the optimal motion vector; E(-1,0) is the cost value of the edge motion vector (-1,0) of the optimal motion vector (0,0) if the optimal motion vector is taken as the center; E(1,0) is the cost value of the edge motion vector (1,0) of the optimal motion vector (0,0) if the optimal motion vector is taken as the center; E(0, -1) is the cost value of the edge vector (0, -1) of motion of the optimal motion vector (0,0) if the optimal motion vector is taken as the center; E(0,1) is the cost value of the edge motion vector (0,1) of the optimal motion vector (0,0) if the optimal motion vector is taken as the center. The method of determining the cost value of each motion vector can be learned from the previous embodiments, and therefore will not be disclosed in detail in this paragraph.

[187] After determining the first sub-pixel motion vector correction value according to the method disclosed above, it is possible to determine the second sub-pixel motion vector correction value based on the first sub-pixel motion vector correction value, wherein the second sub-pixel motion vector correction value is a value symmetric to the first sub-pixel motion vector correction value movement. For example, if the first sub-pixel motion vector correction value is (1, 0), then the second sub-pixel motion vector correction value is (-1, 0), i. value symmetrical to (1, 0).

[188] In step a8, the first source motion vector is corrected according to the first integer motion vector correction value and/or the first subpixel motion vector correction value to obtain the first target motion vector.

[189] For example, the first target motion vector = the first source motion vector + the first integer-pixel motion vector correction value + the first sub-pixel motion vector correction value. Of course, the above example is not limiting.

[190] In step a9, the second source motion vector is corrected according to the second integer motion vector correction value and/or the second subpixel motion vector correction value to obtain the second target motion vector.

[191] For example, the second target motion vector = the second source motion vector + the second integer-pixel motion vector correction value + the second sub-pixel motion vector correction value. Of course, the above example is not limiting.

[192] Embodiment 12: in step 303 and step 403, for each sub-block of the current block, adjust the first source motion vector according to the first pixel value of the first reference block and the second pixel value of the second reference block to obtain the first target motion vector of the sub-block; and adjusting the second source motion vector according to the first pixel value of the first reference block and the second pixel value of the second reference block to obtain a second target motion vector of the subblock.

[193] The original motion vector correction process is disclosed as a procedure for processing one sub-block (eg, each sub-block of size dx*dy of the current block, for example, a sub-block of 16*16 or smaller block). The first source motion vector is designated Org_MV0, the second source motion vector is designated Org_MV1, the first target motion vector obtained after the correction of the first original motion vector Org_MV0 is designated Refined_MV0, and the second target motion vector obtained after the correction of the second source motion vector Org_MV1 is designated Refined_MV1.

[194] In step b1, iterations by the number of SRs are performed to obtain the optimal integer pixel offset of the integer pixel VD point, denoted by IntegerDeltaMV and representing the first integer pixel value of the motion vector correction in the embodiments disclosed above. For example, set the initial value (0, 0) for IntegerDeltaMV, and then perform processing for each iteration as follows:

[195] At step b11 set the value (0, 0) for deltaMV. When the first iteration is performed, the prediction value block A1 (ie, the W*H block at the center of the first reference block) is copied based on the reference pixel of the first original motion vector in the first reference block; the prediction value block B1 (ie, the W*H block at the center of the second reference block) is copied based on the reference pixel of the second original motion vector in the second reference block. The initial cost value is obtained (the initial cost value is the CAP (sum of absolute differences (SAD)) based on the prediction value block A1 and the prediction value block B1) based on the prediction value block A1 and the prediction value block B1, wherein the method for determining the initial cost value is disclosed in the following embodiments. If the initial cost value is less than 4*dx*dy/2, where dx and dy are the width and height of the current sub-block, the subsequent search process is directly skipped, step b2 is performed and notZeroCost is set to false.

[196] At step b12 in Fig. 6, five offset VDs (all referred to as "MVOffset") are sequentially obtained in the sequence {Mv(0.1), Mv(0,-1), Mv(1.0), Mv( -1,0), Mv(right, down)}, with the above starting point taken as the center, and a process of calculating and comparing cost values of said five offset PDs is performed.

[197] For example, based on the determined MVOffset (e.g., Mv(0, 1), etc.) in the first and second reference blocks, two prediction value blocks are obtained (e.g., a W*H block with an MVOffset center offset in the first reference block). block and the W*H block with center position offset -MVOffset (opposite of MVOffset in the second reference block) based on this MVOffset, and downsampling the CAP of said two prediction value blocks as the MVOffset cost value is calculated.

[198] Next, the MVOffset with the minimum cost value is reserved, this MVOffset value is updated to the deltaMV value, and applied as the new center offset point of the next iteration.

[199] The value of IntegerDeltaMV is updated based on deltaMV, with updated IntegerDeltaMV = IntegerDeltaMV before update + deltaMV, i.e. add deltaMV based on the current IntegerDeltaMV.

[200] In step b13, after iteration, if the optimal VD is still the original VD (i.e., not MVOffset) or the minimum cost value is 0, the next iterative search process is not performed, step b2 is performed and set to "false" for notZeroCost. Otherwise, if the number of iterations has reached SR, step b2 is performed, and if the number of iterations has not reached SR, the optimal VD is taken as the center and the next iterative search process is performed, i.e., return to step b11.

[201] After the iterative search process is completed, the IntegerDeltaMV value, namely the final IntegerDeltaMV value, that is, the first integer pixel motion vector correction value, hereinafter referred to as IntegerDeltaMV, is obtained.

[202] In step b2, the optimal point of the whole-pixel VD from step b1 is taken as the center, then the optimal sub-pixel offset VD is obtained, which is referred to as SPMV, wherein SPMV is the first sub-pixel motion vector correction value in the embodiments disclosed above. For example, you can set the initial value (0, 0) for SPMV and perform post-processing as follows.

[203] In step b21, subsequent processing (namely, obtaining SPMV) is performed only if notZeroCost is not "false" and deltaMV is (0, 0), otherwise the original motion vector is corrected directly using IntegerDeltaMV instead of using IntegerDeltaMV and SPMV.

[204] In step b22, E(x, y) is presented as the cost value corresponding to the offset PD of the optimal PD point (x, y) obtained in step b1 (the cost value calculated in step b1). Based on E (x, y) of five points - center, top, bottom, left and right - the offset (x ₀ , y ₀ ) of the point with the smallest E(x, y) can be obtained as follows: x ₀ = N*( E(-1.0) - E(1.0)) / (E(-1.0) + E(1.0) - 2* E(0.0)), y ₀ = N*(E( 0,-1) - E(0,1)) / (E(0,-1) + E(0,1) - 2* E(0,0)).

[205] In one example, if the motion vector pixel precision is 1/2, 1/4, 1/8, and 1/16, then N=1, 2, 4, and 8, respectively. Next, (x ₀ , y ₀ ) is assigned to deltaMv, SPMV = deltaMv/2N, and if the pixel precision of the current motion vector is 1/16, then SPMV is (x ₀ /16, y ₀ /16).

[206] If E(-1, 0) = E(0, 0), shift half a pixel horizontally to the left (deltaMv[0] = -N).

[207] If E(1, 0) = E(0, 0), shift half a pixel horizontally to the right (deltaMv[0] = N).

[208] If E(0, -1) = E(0, 0), perform a shift of half a pixel vertically up (deltaMv [1] = -N).

[209] If E(0, 1) = E(0, 0), perform a shift of half a pixel vertically down (deltaMv [1] = N).

[210] The above processing allows you to get the value of SPMV, ie. the first subpixel motion vector correction value.

[211] In step b3, an optimal offset VD based on IntegrarteltaMV with integer pixel offset in step b1 and SPMV with subpixel offset in step b2 is obtained and the optimal offset VD is designated BestMVoffset. Also, BestMVoffset=IntegerDeltaMV+ SPMV.

[212] Based on BestMVoffset, target motion vectors in two directions can be obtained: Refined_MV0=Org_MV0+BestMVoffset; Refined_MV1=Org_MV1-BestMVoffset.

[213] It should be noted that BestMVoffset=IntegerDeltaMV+SPMV, ie. the sum of the first integer pixel motion vector correction value and the first subpixel motion vector correction value. Also, -IntegerDeltaMV is a value that is symmetrical to IntegerDeltaMV, i.e. the second integer pixel motion vector correction value, and -SPMV is the value symmetrical to SPMV, i.e. second subpixel motion vector correction value, thus -BestMVoffset=(-IntegerDeltaMV)+(-SPMV), i.e. the sum of the second integer pixel motion vector correction value and the second subpixel motion vector correction value.

[214] Embodiment 13: In step 303 and step 403, the implementation of correcting the first and second source motion vectors Org_MV0 and Org_MV1 to the first and second target motion vectors Refined_MV0 and Refined_MV1, respectively, is the same as the implementation in Embodiment 12, except that:

[215] the phrase "if the initial cost value is less than 4*dx*dy/2, ..., the subsequent search process is directly skipped" in step b11 is omitted, that is, even if the initial cost value is less than 4*dx*dy/2, then instead to "directly skip the subsequent search process", the subsequent search process is continued, that is, step b12 must be performed.

[216] Embodiment 14: In step 303 and step 403, the implementation of correcting the first and second source motion vectors Org_MV0 and Org_MV1 to the first and second target motion vectors Refined_MV0 and Refined_MV1, respectively, is the same as the implementation in Embodiment 12, except that:

[217] the phrase "if the initial cost value is less than 4*dx*dy/2, the subsequent search process is directly skipped" in step b11 is omitted, that is, even if the initial cost value is less than 4*dx*dy/2, then instead of "directly skip the subsequent search process", the subsequent search process is continued, that is, step b12 must be performed.

[218] the phrase "if the optimal VP is still the original VP (i.e., not MVOffset) or the minimum cost value is 0, the next iterative search process is not performed" in step b13 is omitted, i.e., even if the optimal VP is still is the original PD or the minimum cost value is 0, the following iterative search process can be performed.

[219] Embodiment 15: In step 303 and step 403, the implementation of correcting the first and second source motion vectors Org_MV0 and Org_MV1 to the first and second target motion vectors Refined_MV0 and Refined_MV1, respectively, is the same as the implementation in Embodiment 12, except that:

[220] The "notZeroCost" correlation process is omitted, that is, notZeroCost is not set and stored in step b11 and step b13. In step b21, it is possible to execute the sub-pixel offset calculation process (i.e., step b22) if deltaMV is (0, 0), not only if notZeroCost is not "false" but deltaMV is (0) , 0).

[221] Embodiment 16: In step 303 and step 403, the implementation of correcting the first and second source motion vectors Org_MV0 and Org_MV1 to the first and second target motion vectors Refined_MV0 and Refined_MV1, respectively, is the same as the implementation in Embodiment 12, except that:

[222] the phrase "post-processing is performed only if notZeroCost is not false and deltaMV is (0, 0), otherwise the original motion vector is corrected directly using IntegerDeltaMV" in step b21 is modified to " post-processing is performed only if notZeroCost is not "false", while the cost values at four points located respectively above, below, left and right at a distance of one pixel from the current optimal integer pixel are obtained by calculation in step b1, otherwise, the original motion vector is corrected directly using IntegerDeltaMV."

[223] In one example, "post-processing" means the process of calculating the sub-pixel offset in step b22.

[224] In one example, in the process of calculating the sub-pixel offset in step b22, it is necessary to apply cost values at four points located respectively above, below, left, and right at a distance of one pixel from the current optimal integer pixel, that is, calculating “four-point cost values located respectively above, below, left and right at a distance of one pixel from the current optimal integer pixel" in step b1 is a prerequisite.

[225] Embodiment 17: In step 303 and step 403, the implementation of correcting the first and second source motion vectors Org_MV0 and Org_MV1 to the first and second target motion vectors Refined_MV0 and Refined_MV1, respectively, is the same as the implementation in Embodiment 12, except that:

[226] the phrase "post-processing is performed only if notZeroCost is not false and deltaMV is (0, 0), otherwise the original motion vector is corrected directly using IntegerDeltaMV" in step b21 is modified to " the post-processing (i.e., the sub-pixel offset calculation process) is performed only if the cost values at four points located respectively above, below, left, and right at a distance of one pixel from the current optimal integer pixel are obtained by calculation in step b1, otherwise, the original motion vector is corrected directly using IntegerDeltaMV."

[227] Embodiment 18: In step 303 and step 403, the implementation of correcting the first and second source motion vectors Org_MV0 and Org_MV1 to the first and second target motion vectors Refined_MV0 and Refined_MV1, respectively, is the same as the implementation in Embodiment 12, except that:

[228] the phrase "post-processing is performed only if notZeroCost is not false and deltaMV is (0, 0), otherwise the original motion vector is corrected directly using IntegerDeltaMV" in step b21 is modified to " the subsequent processing (i.e., the sub-pixel offset calculation process in step b22) is performed if the cost values at four points located respectively above, below, left, and right at a distance of one pixel from the current optimal integer pixel are obtained in step b1, otherwise case, perform step b23".

[229] At step b23, the nearest integer pixel point MV_inter_nearest is taken as the current optimal integer pixel point MV_inter_org, while the cost values at four points located respectively above, below, left and right at a distance of one pixel from the nearest integer pixel point were obtained at step b1 . Next, the sub-pixel offset calculation process in step b22 is performed centered on MV_inter_nearest, that is, an SPMV centered on MV_inter_nearest is obtained.

[230] For example, if the cost values at four points located respectively above, below, left and right at a distance of one pixel from the current optimal integer pixel point were not obtained in full in step b1, one integer MV_inter_nearest point is selected from the area near the border the optimal integer point MV_inter_org, while the cost values at four points located respectively above, below, left and right of the integer pixel point MV_inter_nearest are obtained in step b1.

[231] Next, the integer pixel point MV_inter_nearest is taken as the current optimal integer pixel point, and an SPMV centered on the integer pixel point MV_inter_nearest is obtained, with the specific process related to step b22.

[232] In the process of obtaining SPMV centered on the integer pixel point MV_inter_nearest in step b22, the calculation of x ₀ and y ₀ may be limited to the range of [-2N, 2N]. If x ₀ /y ₀ is greater than 2N, x ₀ /y ₀ can be assigned the value 2N; if x ₀ /y ₀ is less than -2N, x ₀ /y ₀ can be assigned the value -2N. If the motion vector pixel precision is 1/2, 1/4, 1/8, and 1/16, then N=1, 2, 4, and 8, respectively.

[233] Embodiment 19: In the embodiments disclosed above, it is necessary to determine edge motion vectors corresponding to the center motion vector, for example, 5 edge motion vectors. For example, the central motion vector (x, y) can be shifted by S in different directions and sequentially receive the edge vector (x, y+S) of the motion, the edge vector (x, y-S) of the motion, the edge vector (x+S, y) of the motion , edge vector (x-S, y) of movement and edge vector (x+right, y+down) of movement in different directions. Alternatively, the center motion vector (x, y) can be shifted by S in different directions and successively obtain edge motion vector (x, y-S), edge vector (x, y+S) motion, edge vector (x-S, y) motion , edge vector (x+S, y) of movement and edge vector (x + right, y + down) of movement in different directions.

[234] For example, if (x, y) is (0, 0) and S is 1, then you can get 5 edge motion vectors in the sequence (0, 1), (0, -1), (1, 0 ), (-1, 0) and (right, down). Alternatively, you can get 5 edge motion vectors in the sequence (0, -1), (0, 1), (-1, 0), (1, 0) and (right, down).

[235] Embodiment 20: In the embodiments disclosed above, the default value of the edge vector (x+right, y+down) of motion is (x-S, y-S). If the cost value of the edge vector (x+S, y) of motion is less than the value of the edge vector (x-S, y) of motion, then the value of "right" is S (-S is changed to S); if the cost value of the edge vector (x, y+S) of the motion is less than the value of the edge vector (x, y-S) of the motion, then the "down" value is S (-S is changed to S). Alternatively, if the cost value of the edge vector (x+S, y) of the motion is less than or equal to the value of the cost of the edge vector (x-S, y) of the motion, the value of "right" is S (-S is changed to S); if the cost value of the edge vector (x, y+S) of the motion is less than or equal to the cost value of the edge vector (x, y-S) of the motion, the "down" value is S (-S is changed to S).

[236] For example, 5 edge motion vectors are obtained in the sequence (0, 1), (0, -1), (1, 0), (-1, 0), (right, down), with the default value (right, down) is (-1, -1). If the cost value of the edge vector (1, 0) of the motion is less than the value of the cost of the edge vector (-1, 0) of the motion, the value of "right" is 1; if the value of the edge vector (0, 1) motion cost is less than the value of the edge vector (0, -1) motion, the down value is 1. Alternatively, if the value of the edge vector (1, 0) motion vector is less than or equal to the value the cost of the edge vector (-1, 0) movement, the value of "right" is 1; if the edge vector cost value (0, 1) of the motion is less than or equal to the edge vector cost value (0, -1) of the motion, the down value is 1.

[237] For example, 5 edge motion vectors are obtained in the sequence (0, -1), (0, 1), (-1, 0), (1, 0), (right, down), with the default value (right, down) is (-1, -1). If the cost value of the edge vector (1, 0) of the motion is less than the value of the cost of the edge vector (-1, 0) of the motion, the value of "right" is 1; if the value of the edge vector (0, 1) motion cost is less than the value of the edge vector (0, -1) motion, the down value is 1. Alternatively, if the value of the edge vector (1, 0) motion vector is less than or equal to the value the cost of the edge vector (-1, 0) movement, the value of "right" is 1; if the edge vector cost value (0, 1) of the motion is less than or equal to the edge vector cost value (0, -1) of the motion, the down value is 1.

[238] Embodiment 21: in steps 303 and 403, for each subblock of the current block, adjust the first source motion vector according to the first pixel value of the first reference block and the second pixel value of the second reference block to obtain the first target motion vector of the subblock; while correcting the second source motion vector according to the first pixel value of the first reference block and the second pixel value of the second reference block to obtain the second target motion vector of the subblock.

[239] The original motion vector correction process is disclosed as a procedure for processing one sub-block (eg, each sub-block of size dx*dy of the current block, for example, a sub-block of 16*16 or smaller block).

[240] In step c1, the first source motion vector or the second source motion vector is taken as the center, some or all of the motion vectors are selected from the surrounding motion vectors, and the selected motion vectors are determined as candidate motion vectors.

[241] For example, some or all of the motion vectors can be selected from the surrounding motion vectors as potential motion vectors, with the first original motion vector being taken as the center. Details are disclosed in the following embodiments.

[242] In the following embodiments, the first original motion vector is taken as the center motion vector for convenience of disclosure.

[243] In step c2, a third cost value corresponding to the first original motion vector and a fourth cost value corresponding to each potential motion vector are obtained according to the first pixel value of the first reference block and the second pixel value of the second reference block.

[244] For example, the reference subblock A1 corresponding to the first original motion vector is obtained by copying from the first reference block, while the reference subblock A1 is the reference subblock of the first original motion vector in the first reference block, while the size of the reference subblock A1 is the size current block. Reference sub-block B1 corresponding to the second original motion vector is obtained by copying from the second reference block, wherein reference sub-block B1 is the reference sub-block of the second original motion vector in the second reference block, and the size of reference sub-block B1 is the size of the current block. Next, a third cost value corresponding to the first original motion vector is obtained according to the first pixel value of the reference sub-block A1 and the second pixel value of the reference sub-block B1.

[245] For each potential motion vector, a reference sub-block A2 corresponding to the potential motion vector is obtained by copying from the first reference block, wherein the reference sub-block A2 is the reference sub-block of the potential motion vector in the first reference block, while the size of the reference sub-block A2 is the size of the current block. The B2 reference subblock corresponding to the motion vector symmetrical to the potential motion vector is obtained by copying from the second reference block, wherein the size of the B2 reference subblock is the size of the current block. A fourth cost value corresponding to the potential motion vector is obtained according to the first pixel value of the reference sub-block A2 and the second pixel value of the reference sub-block B2.

[246] In step c3: a motion vector is selected from the first original motion vector and potential motion vectors as the optimal motion vector based on the third cost value and the fourth cost value. For example, the motion vector with the lowest cost value (eg, the first original motion vector or any of the potential motion vectors) is used as the optimal motion vector.

[245] In step c4, a first integer pixel motion vector correction value (which is used to correct the first original motion vector) and a second integer pixel motion vector correction value (which is used to correct the second original motion vector) are determined based on the optimal motion vector. In one example, the first integer pixel motion vector correction value may be determined based on the optimal motion vector and the first original motion vector, and the second integer pixel motion vector correction value may be determined according to the first integer pixel motion vector correction value, wherein the second integer pixel motion vector correction value is symmetrical to the first integer pixel value. motion vector correction.

[248] For example, if the optimal motion vector has coordinates (4, 6) and the first original motion vector has coordinates (4, 4), then the first integer pixel motion vector correction value is determined based on the optimal motion vector (4, 6) and the first of the original motion vector (4, 4), wherein the first integer value of the motion vector correction is the difference between the optimal motion vector (4, 6) and the first original motion vector (4, 4), i.e. the first integer pixel value of the motion vector correction is (0, 2).

[249] Next, a second integer pixel motion vector correction value is determined according to the first integer pixel motion vector correction value (0, 2), wherein the second integer pixel motion vector correction value may be (0, -2), i. be a value symmetric to (0, 2).

[250] In step c5, a first sub-pixel motion vector correction value (which is used to correct the first original motion vector) and a second sub-pixel motion vector correction value (which is used to correct the second original motion vector) are determined based on the optimal motion vector.

[251] In one example, the first sub-pixel motion vector correction value can be determined based on the cost value corresponding to the optimal motion vector and the cost value corresponding to the edge motion vector corresponding to the optimal motion vector, and then the second sub-pixel motion vector correction value can be determined based on the first sub-pixel motion vector correction value.

[252] For example, x ₀ =N*(E(-1.0) - E(1.0)) / (E(-1.0) + E(1.0) - 2* E(0.0 )), y ₀ =N*(E(0,-1) - E(0.1)) / (E(0,-1) + E(0.1) - 2* E(0.0)) , while if the pixel accuracy of the motion vector is 1/2, 1/4, 1/8 and 1/16, then N =1, 2, 4 and 8, respectively. Next, (x ₀ , y ₀ ) is assigned to deltaMv, SPMV = deltaMv/2N, whereby if the pixel precision of the current motion vector is 1/16, then SPMV is (x ₀ /16, y ₀ /16).

[253] SPMV is the first subpixel motion vector correction value. E(0,0) - value of the cost of the optimal motion vector; E(-1,0) - the value of the cost of the edge vector (-1,0) motion of the optimal motion vector (0,0) if the optimal motion vector is taken as the center; E(1,0) - value of the cost of the edge vector (1,0) motion of the optimal motion vector (0,0) if the optimal motion vector is taken as the center; E(0, -1) - value of the cost of the edge vector (0, -1) of the movement of the optimal motion vector (0,0) if the optimal motion vector is taken as the center; E(0,1) is the value of the cost of the edge vector (0,1) of the optimal motion vector (0,0) of the motion, if the optimal motion vector is taken as the center. The method for determining the cost value of each motion vector can be learned from the embodiments disclosed above.

[254] After determining the first sub-pixel motion vector correction value according to the method disclosed above, it is possible to determine the second sub-pixel motion vector correction value based on the first sub-pixel motion vector correction value, wherein the second sub-pixel motion vector correction value is a value symmetric to the first sub-pixel motion vector correction value movement. For example, if the first sub-pixel motion vector correction value is (1, 0), then the second sub-pixel motion vector correction value is (-1, 0), i. is a value symmetric to (1, 0).

[255] In step c6, the first source motion vector is corrected according to the first integer motion vector correction value and/or the first subpixel motion vector correction value to obtain the first target motion vector.

[256] For example, the first target motion vector = the first source motion vector + the first integer-pixel motion vector correction value + the first sub-pixel motion vector correction value. Of course, the above example is not limiting.

[257] In step c7, the second source motion vector is corrected according to the second integer motion vector correction value and/or the second subpixel motion vector correction value to obtain the second target motion vector.

[258] For example, the second target motion vector = the second source motion vector + the second integer-pixel motion vector correction value + the second sub-pixel motion vector correction value. Of course, the above example is not limiting.

[259] Embodiment 22: in step 303 and step 403, for each sub-block of the current block, adjust the first source motion vector according to the first pixel value of the first reference block and the second pixel value of the second reference block to obtain the first target motion vector of the sub-block; and adjusting the second source motion vector according to the first pixel value of the first reference block and the second pixel value of the second reference block to obtain a second target motion vector of the subblock.

[260] The original motion vector correction process is disclosed as a procedure for processing one sub-block (eg, each sub-block of size dx*dy of the current block, for example, a sub-block of size 16*16, or a smaller block). The first source motion vector is denoted Org_MV0, the second source motion vector is denoted Org_MV1, the first target motion vector is denoted Refined_MV0, and the second target motion vector is denoted Refined_MV1.

[261] In step d1, some or all of the motion vectors are selected from (2*SR+1)*(2*SR+1) points around the first original motion vector, with the first original motion vector taken as the center. For example, if SR=2, then some or all of the 25-point motion vectors around the first original motion vector are selected as candidate motion vectors. Next, the value of the cost of the first initial motion vector is determined and the value of the cost of each potential motion vector is determined. Next, the motion vector with the minimum cost value is used as the optimal motion vector.

[262] In contrast to step b1 of the above embodiment, in step d1, all potential motion vectors to be processed can be selected simultaneously without performing an iterative process, instead of selecting some motion vectors during the first iteration and selecting other motion vectors during the second iteration. iterations of an iterative process. Due to this, all potential motion vectors to be processed are selected simultaneously so that the potential motion vectors can be processed in parallel to obtain a cost value of each potential motion vector, which can reduce computational complexity and improve coding performance.

[263] In step d2, an IntegerDeltaMV value is determined based on the optimal motion vector, the final IntegerDeltaMV value being the first integer pixel value of the motion vector correction; the method of determination will not be rediscovered in this paragraph.

[264] In step d3: the optimal motion vector is taken as the center, the optimal sub-pixel offset VD is obtained, which is denoted SPMV, wherein SPMV is the first sub-pixel motion vector correction value.

[265] The implementation process of step d3 can be found above in the description of step b2, so it will not be disclosed again in this paragraph.

[266] In step d4, a BestMVoffset is obtained based on IntegerDeltaMV and SPMV. For example, BestMVoffset= IntegerDeltaMV+SPMV.

[267] The target motion vector is obtained based on the BestMVoffset: Refined_MV0=Org_MV0+BestMVoffset; Refined_MV1=Org_MV1-BestMVoffset.

[268] Embodiment 23: In step 303 and step 403, the implementation of correcting the first and second source motion vectors Org_MV0 and Org_MV1 to the first and second target motion vectors Refined_MV0 and Refined_MV1, respectively, is similar to the implementation in Embodiments 21, 22.

[269] In this embodiment, all motion vectors are selected from (2*SR+1)*(2*SR+1) points around the original motion vector, with the original motion vector taken as the center. For example, if SR=2, then all motion vectors are selected from 25 points around the original motion vector, the cost values of these vectors are determined, and the cost value of each motion vector is determined. Next, the motion vector with the minimum cost value is used as the optimal motion vector.

[270] Embodiment 24: In step 303 and step 403, the implementation of correcting the first and second source motion vectors Org_MV0 and Org_MV1 to the first and second target motion vectors Refined_MV0 and Refined_MV1, respectively, is similar to the implementation in Embodiments 21, 22. In this embodiment, all subjects processing, the potential motion vectors are selected simultaneously so that the potential motion vectors can be processed in parallel to obtain a cost value for each potential motion vector, which can reduce computational complexity and improve coding performance.

[271] In this embodiment, some of the motion vectors with an offset of no more than SR are selected from (2*SR+1)*(2*SR+1) points around the original motion vector, with the original motion vector taken as the center. For example, N (N is not less than 1 and not more than (2*SR+1)*(2*SR+1)) potential points are selected from (2*SR+1)*(2*SR+1) points, including the original motion vector. The cost values of the motion vectors corresponding to the specified N points are determined. The cost values of N points are viewed in a certain sequence and the motion vector with the minimum cost value is selected as the optimal motion vector. If the cost values are equal, the highest order potential points are preferably selected.

[272] In one example, if SR=2, the number of candidate points may be 25, with the sequence of candidate points being left to right and top to bottom. 7A, the sequence of potential points may be: { Mv(-2, -2), Mv(-1, -2), Mv(0, -2), Mv(1, -2), Mv(2, - 2), Mv(-2, -1), Mv(-1, -1), Mv(0, -1), Mv(1, -1), Mv(2, -1), Mv(-2, 0), Mv(-1, 0), Mv(0, 0), Mv(1, 0), Mv(2, 0), Mv(-2, 1), Mv(-1, 1), Mv( 0, 1), Mv(1, 1), Mv(2, 1), Mv(-2, 2), Mv(-1, 2), Mv(0, 2), Mv(1, 2), Mv (2, 2) }.

[273] The cost values corresponding to the motion vectors of 25 points are determined, scanned in the above sequence, a motion vector with a minimum cost value is obtained in the form of an optimal offset VD, with which an integer pixel motion vector correction value and a subpixel motion vector correction value can be determined, when this determination methods can be learned from the embodiments disclosed above.

[274] In another example, if SR=2, then the number of candidate points may be 21, with the sequence of candidate points being left to right and top to bottom. On FIG. 7B the sequence of potential points can be: {Mv(-1,-2), Mv(0,-2), Mv(1,-2), Mv(-2,-1), Mv(-1,-1) , Mv(0,-1), Mv(1, -1), Mv(2, -1), Mv(-2, 0), Mv(-1, 0), Mv(0, 0), Mv( 1, 0), Mv(2, 0), Mv(-2, 1), Mv(-1, 1), Mv(0, 1), Mv(1, 1), Mv(2, 1), Mv (-1, 2), Mv(0, 2), Mv(1, 2)}. The cost values corresponding to the motion vectors of 21 points are determined, the view is performed in the above sequence, the motion vector with the minimum cost value is obtained in the form of an optimal offset VD, with which it is possible to determine the integral pixel value of the motion vector correction and the subpixel value of the motion vector correction, while methods for determining can be learned from the embodiments disclosed above.

[275] In another example, if SR = 2, the number of candidate points may be 25, with the sequence of these candidate points being a sequence from an area near the center to an area away from it, with the (0, 0) motion vector taken as center. In Fig.7C, the sequence of potential points can be: {Mv(0, 0) , Mv(-1, 0), Mv(0,-1), Mv(1, 0), Mv(0, 1), Mv( -1, 1), Mv(-1,-1), Mv(1,-1), Mv(1, 1), Mv(0, 2), Mv(-2, 0), Mv(0,- 2), Mv(2, 0), Mv(1, 2), Mv(-1, 2), Mv(-2, 1), Mv(-2,-1), Mv(-1,-2) , Mv(1,-2), Mv(2,-1), Mv(2, 1), Mv(-2, 2), Mv(-2,-2), Mv(2,-2), Mv (2, 2)}. The cost values corresponding to the motion vectors of 25 points are determined, the scan is performed in the above sequence, the motion vector with the minimum cost value is obtained in the form of an optimal shifted VD, with which you can determine the integral pixel value of the motion vector correction and the subpixel value of the motion vector correction, while methods for determining can be learned from the embodiments disclosed above.

[276] In another example, if SR =2, the number of potential points may be 21, with the sequence of these potential points being a sequence from an area near the center to an area away from it, with the motion vector (0, 0) taken as center. 7D, the sequence of potential points is: { Mv(0.0), Mv(-1.0), Mv(0, -1), Mv(1.0), Mv(0.1), Mv( -1.1), Mv(-1, -1), Mv(1, -1), Mv(1.1), Mv(0.2), Mv(-2.0), Mv(0, - 2), Mv(2,0), Mv(1, 2), Mv(-1,2), Mv(-2,1), Mv(-2, -1), Mv(-1, -2) , Mv(1, -2), Mv(2, -1), Mv(2, 1) }.

[277] The cost values corresponding to the motion vectors of the 21 points are determined, scanned in the above sequence, the motion vector with the minimum cost value is obtained in the form of the optimal offset VD, with which the integral pixel motion vector correction value and the subpixel motion vector correction value can be determined, when this determination methods can be learned from the embodiments disclosed above.

[278] In another example, if SR =2, the number of potential points may be 13, with the sequence of these potential points being a sequence from an area near the center to an area away from it, with the motion vector (0, 0) taken as center. 7E, the sequence of potential points is: {Mv(0.0), Mv(-1.0), Mv(0,-1), Mv(1.0), Mv(0.1), Mv( -1,1), Mv(-1,-1), Mv(1,-1), Mv(1, 1), Mv(0, 2), Mv(-2, 0), Mv(0,- 2), Mv(2, 0)}. The cost values corresponding to the motion vectors of 13 points are determined, the view is performed in the above sequence, the motion vector with the minimum cost value is obtained in the form of an optimal offset VD, with which it is possible to determine the integral pixel value of the motion vector correction and the subpixel value of the motion vector correction, while methods for determining can be learned from the embodiments disclosed above.

[279] Embodiment 25: In step 303 and step 403, the implementation of correcting the first and second source motion vectors Org_MV0 and Org_MV1 to the first and second target motion vectors Refined_MV0 and Refined_MV1, respectively, is similar to the implementation in Embodiments 21, 22. In this embodiment, all subjects processing, the potential motion vectors are selected simultaneously so that the potential motion vectors can be processed in parallel to obtain a cost value for each potential motion vector, which can reduce computational complexity and improve coding performance.

[280] In this embodiment, some of the motion vectors with an offset of no more than SR are selected from (2*SR+1)*(2*SR+1) points around the original motion vector, with the original motion vector taken as the center. For example, N (N is not less than 1 and not more than (2*SR+1)*(2*SR+1)) potential points are selected from (2*SR+1)*(2*SR+1) points, including the original motion vector. The cost values of the motion vectors corresponding to the specified N points are determined. The cost values of N points are viewed in a certain sequence and the motion vector with the minimum cost value is selected as the optimal motion vector. If the cost values are equal, the highest order potential points are preferably selected.

[281] Embodiment 25 differs from Embodiment 24 in that all potential dot positions of Embodiment 24 are fixed, i. independent of the original motion vector, while the positions of the potential points in embodiment 25 are associated with the original motion vector, a few particular examples of which are described below.

[282] In one example, if SR =2, then the number of potential points may be 13, with the sequence of these potential points being a sequence from the area near the center to the area away from it, with the motion vector (0, 0) taken in as a centre. In addition, the sequence of potential points of the first layer from the center does not depend on the size of the original motion vector, and the sequence of potential points of the second layer from the center depends on the size of the original motion vector. The sequence of potential points is: {Mv(0.0), Mv(-1.0), Mv(0,-1), Mv(1.0), Mv(0.1), Mv(-1.1 ), Mv(-1,-1), Mv(1,-1), Mv(1,1), Mv(sign_H*2,0), Mv(sign_H*2,sign_V*1), Mv(sign_H* 1, sign_V *2), Mv(sign_H*1, sign_V *2)}. The first original motion vector is designated MV0, the horizontal component is MV0_Hor, and the vertical component is MV0_Ver. If MV0_Hor is not less than 0, then sign_H = 1; otherwise sign_H = -1; if MV0_Ver is not less than 0, then sign_V = 1; otherwise sign_V = -1.

[283] The cost values corresponding to the motion vectors of the 13 points are determined, scanned in the above sequence, the motion vector with the minimum cost value is obtained in the form of the optimal offset VD, with which the integral pixel motion vector correction value and the subpixel motion vector correction value can be determined, when this determination methods can be learned from the embodiments disclosed above.

[284] In another example, if SR =2, then the number of potential points may be 13, with the sequence of these potential points being a sequence from the area near the center to the area away from it, with the motion vector (0, 0) taken in as a centre. In addition, the sequence of potential points of the first layer from the center does not depend on the size of the original motion vector, and the sequence of potential points of the second layer from the center depends on the size of the original motion vector. The sequence of potential points is: {Mv(0.0), Mv(-1.0), Mv(0,-1), Mv(1.0), Mv(0.1), Mv(-1.1 ), Mv(-1,-1), Mv(1,-1), Mv(1,1), Mv(sign_H*2,0), Mv(sign_H*2,sign_V*1), Mv(sign_H* 1, sign_V *2), Mv(sign_H*1, sign_V *2)}. The first original motion vector is designated MV0, the horizontal component is MV0_Hor, and the vertical component is MV0_Ver. If MV0_Hor is greater than 0, then sign_H = 1; otherwise sign_H = -1; if MV0_Ver is greater than 0, then sign_V = 1; otherwise sign_V = -1.

[285] The cost values corresponding to the motion vectors of the 13 points are determined, scanned in the above sequence, the motion vector with the minimum cost value is obtained in the form of the optimal offset VD, with which the integral pixel motion vector correction value and the subpixel motion vector correction value can be determined, when this determination methods can be learned from the embodiments disclosed above.

[286] Embodiment 26: In the embodiments disclosed above, the first cost value corresponding to the center motion vector and the second cost value corresponding to the edge motion vector are obtained based on the first pixel value of the first reference block and the second pixel value of the second reference block. A third cost value corresponding to the first original motion vector and a fourth cost value corresponding to the potential motion vectors are obtained based on the first pixel value of the first reference block and the second pixel value of the second reference block.

[287] In one example, a first cost value corresponding to a center motion vector, a second cost value corresponding to an edge motion vector, a third cost value corresponding to a first source motion vector, and a fourth cost value corresponding to a potential motion vector can be obtained based on the first value. a non-downsampling pixel and a second non-downsampling pixel value.

[288] In another example, a downsampling operation is performed on the first and second pixel values; further, based on the first and second pixel values after downsampling, the first cost value corresponding to the central motion vector, the second cost value corresponding to the edge motion vector, the third cost value corresponding to the first original motion vector, and the fourth cost value corresponding to the potential motion vector are obtained.

[289] In another example, a downsampling and shifting operation is performed on the first and second pixel values, and then based on the first and second pixel values after shifting and downsampling, a first cost value corresponding to the central motion vector, a second cost value corresponding to the edge motion vector, a third cost value corresponding to the first original motion vector, and a fourth cost value corresponding to the potential motion vector.

[290] The method including determining the cost value is similar for different situations. For example, to obtain a cost value corresponding to the central motion vector, the reference sub-block A1 corresponding to the central motion vector may be obtained by copying from the first reference block, and the reference sub-block B1 corresponding to the motion vector symmetric to the central motion vector may be obtained by copying from the second of the reference block, wherein the cost value corresponding to the center motion vector can be obtained according to the first pixel value of the reference subblock A1 and the second pixel value of the reference subblock B1. To obtain a cost value corresponding to an edge motion vector, a reference subblock A2 corresponding to the edge motion vector may be obtained by copying from the first reference block, and a reference subblock B2 corresponding to a motion vector symmetric to the edge motion vector may be obtained by copying from the second reference block. , wherein the cost value corresponding to the edge motion vector can be obtained according to the first pixel value of the reference sub-block A2 and the second pixel value of the reference sub-block B2, and the like.

[291] Thus, in order to obtain a cost value corresponding to a motion vector, a reference subblock corresponding to the motion vector is obtained based on the first reference block, a reference subblock corresponding to a motion vector symmetric to said motion vector is obtained based on the second reference block, and then obtaining a cost value corresponding to the motion vector from the pixel values of said two reference subblocks.

[292] Embodiment 27: Based on Embodiment 26, a cost value corresponding to a motion vector is obtained based on the first non-downsampling pixel value (i.e., the non-downsampling pixel value for the reference subblock in the first reference block) and the second value non-downsampling pixel (ie, non-downsampling pixel values for a reference subblock in the second reference block).

[293] In one example, suppose the reference subblock in the first reference block is pred₀, and the pivot subblock in the second pivot block is pred_one, the cost value is determined based on the CAP of all pixel values of the reference subblocks pred₀, predict_one, and there is no need to perform vertical downsampling on the pixels of the reference subblocks pred₀, predict_one.

[294] Based on all the pixel values of the reference subblocks pred ₀ , pred ₁ , the cost value calculation formula is:

[295] In the above equation, "cost" is the cost value, W is the width value of the reference subblock, H is the height value of the reference subblock, pred ₀ (i,j) is the pixel value of the vertical row i and horizontal row j of the reference subblock pred ₀ , pred ₁ (i,j) is the pixel value of the vertical row i and horizontal row j of the reference subblock pred ₁ , and abs(x) represents the absolute value of x.

[296] Embodiment 28: Based on Embodiment 26, a downsampling operation can be performed on the first and second pixel values; the cost value corresponding to the motion vector is obtained based on the first downsampled pixel value (i.e., the downsampled pixel value for the reference subblock in the first reference block) and the second downsampled pixel value (i.e., the downsampled pixel value sampling for the reference subblock in the second reference block).

[297] In one example, suppose that the reference subblock in the first reference block is pred ₀ and the reference subblock in the second reference block is pred ₁ , the cost value is determined based on the CAP of all pixel values of the reference subblocks pred ₀ , pred ₁ . If the cost value is determined based on the CAP of all pixel values, a vertical downsampling operation can be performed on the pixel values of the reference subblocks pred ₀ , pred ₁ by N times (N is an integer greater than 0 and may be 2).

[298] Based on all the pixel values of the reference subblocks pred ₀ , pred ₁ , the cost value calculation formula is:

[299] In the above equation, "cost" is a cost value, W is a reference subblock width value, H is a reference subblock height value; N represents the downsampling parameter and is an integer greater than 0, which may be 2; pred ₀ (1+N(i-1), j) - pixel value of vertical row 1+N(i-1) and horizontal row j of reference subblock pred ₀ , pred ₁ (1+N(i-1), j) is the pixel value of the vertical row 1+N(i-1) and the horizontal row j of the reference subblock pred ₁ , and abs(x) represents the absolute value of x.

[300] Embodiment 29: Based on Embodiment 26, a downsampling operation is performed on the first and second pixel values, then based on the first and second pixel values (the pixel value after shifting and downsampling for the reference subblock in the first reference block and the pixel value after shifting and downsampling for the reference subblock in the second reference block) after shifting and downsampling, cost values corresponding to motion vectors are obtained.

[301] In one example, suppose that the reference subblock in the first reference block is pred ₀ , the reference subblock in the second reference block is pred ₁ , with both pred ₀ and pred ₁ stored in the form of D bits, that is, each value pixel in pred ₀ is stored in D bits and each pixel value in pred ₁ is stored in D bits.

[302] If D is not greater than 8, then the cost value is determined based on the CAP of all pixel values of the reference subblocks pred ₀ , pred ₁ . If the cost value is determined based on the CAP of all pixel values, a vertical downsampling operation can be performed on the pixel values of the reference subblocks pred ₀ , pred ₁ by N times (N is an integer greater than 0 and may be 2). Based on all the pixel values of the reference subblocks pred ₀ , pred ₁ , the cost value calculation equation is:

[303] In the above equation, "cost" is a cost value, W is a reference subblock width value, H is a reference subblock height value; N represents the downsampling parameter and is an integer greater than 0, which may be 2; pred ₀ (1+N(i-1), j) - pixel value of vertical row 1+N(i-1) and horizontal row j of reference subblock pred ₀ , pred ₁ (1+N(i-1), j) is the pixel value of the vertical row 1+N(i-1) and the horizontal row j of the reference subblock pred ₁ , and abs(x) represents the absolute value of x. From the foregoing, it can be understood that only the sum of the absolute values of the difference in the horizontal row 1, N +1, 2N +1, ... is calculated.

[304] If D is greater than 8, all pixel values of the reference subblocks pred ₀ , pred ₁ are first shifted by 8 bits to obtain 8-bit pred ₀ and 8-bit pred ₁ , denoted by pred _{0_8bit(i, j)} and pred _{1_8bit(i , j)} . The goal is to reduce the cost of storing CAP calculation data, and storing 8 bits can increase the degree of parallelism in data processing.

[305] Next, a vertical downsampling operation can be performed on the pixel values of 8-bit pred ₀ and 8-bit pred ₁ N times (N is an integer greater than 0 and may be 2), which allows the cost value to be calculated as follows:

[306] The meanings of the expressions in the above equation are given in the embodiments disclosed above and will not be repeated in this paragraph.

[307] Embodiment 30: In step 304 and step 404, for each subblock of the current block, motion compensation is performed on the subblock based on the subblock's first target motion vector and the subblock's second target motion vector. Specifically, a third reference block corresponding to a subblock is determined based on the first target motion vector of the subblock, and a fourth reference block corresponding to the subblock is determined based on the second target motion vector of the subblock. The third pixel value of the third reference block and the fourth pixel value of the fourth reference block are weighted to obtain a sub-block prediction value.

[308] For example, reference blocks (i.e., third and fourth reference blocks including 3-component prediction values, and interpolation is necessary because the target motion vector may be sub-pixel) in two directions are obtained by interpolation (for example, using 8-tap interpolation filter) based on the two target motion vectors of the subblock. The third pixel value of the third reference block and the fourth pixel value of the fourth reference block are weighted to obtain a final prediction value (three components).

[309] In one example, a fifth reference block corresponding to a subblock may be determined from the first reference frame based on the first target motion vector of the subblock, and a third reference block may be obtained by interpolating pixel values in the fifth reference block. The sixth reference block corresponding to the subblock may be determined from the second reference frame based on the second target motion vector of the subblock, and the fourth reference block may be obtained by interpolating pixel values in the sixth reference block.

[310] For example, suppose the size of the current block is W*H, the fifth reference block of size A*B can be determined from the first reference frame based on the first target motion vector, the size A*B of the fifth reference block depends on the interpolation method, and, without any limitation, A is greater than W and B is greater than H. A third reference block of size W*H can be obtained by interpolating pixel values in the fifth reference block, and the interpolation method is not limited in this case. The sixth reference block of size A*B can be determined from the second reference frame based on the second target motion vector, while the size A*B of the sixth reference block depends on the interpolation method, where A is greater than W and B is greater than H. The fourth reference block is W*H can be obtained by interpolating the pixel values in the sixth reference block, and the method of interpolation in this case is not limited.

[311] In another example, a seventh reference block corresponding to a subblock is determined from the first reference frame based on the first target motion vector of the subblock, an eighth reference block is built based on the pixel values in the seventh reference block, and the pixel values in the eighth reference block are interpolated to obtain a third reference block. block. The ninth reference block corresponding to the sub-block is determined from the second reference frame based on the second target motion vector of the sub-block, the tenth reference block is built based on the pixel values in the ninth reference block, and the pixel values in the tenth reference block are interpolated to obtain the fourth reference block.

[312] For example, suppose that the size of the current block is W*H, while the seventh reference block of size W*H can be determined from the first reference frame based on the first target motion vector. The eighth reference block of size A*B is built based on the pixel value in the seventh reference block, and the construction method is not limited. The size A*B of the eighth reference block depends on the interpolation method, and without any limitation, A is greater than W and B is greater than H. Pixel values in the eighth reference block are interpolated to obtain a third reference block of size W*H, and the interpolation method is not limited.

[313] For example, suppose that the size of the current block is W*H, while the ninth reference block of size W*H can be determined from the second reference frame based on the second target motion vector. The tenth reference block of size A*B is built based on the pixel value in the ninth reference block, and the construction method is not limited. The size A*B of the tenth reference block depends on the interpolation method, and without any limitation, A is greater than W and B is greater than H. Pixel values in the tenth reference block are interpolated to obtain a fourth reference block of size W*H, and the interpolation method is not limited.

[314] Embodiment 31: After obtaining the target motion vector, based on the target motion vector of each sub-block, bi-directional prediction values (i.e., three components Y, U, V, i.e., the prediction value of the third reference block and the value reference block prediction) with an 8-tap interpolation filter and weight them to obtain final prediction values. On FIG. 5, the pixel values of the black region and the white region are obtained from the reference frame, and the pixel values of the gray region are optionally obtained from the reference frame, and can be obtained by copying adjacent pixel values.

[315] In one example, first W+FS-1 pixel values in the first horizontal row of the white area can be copied to pixel values in the first SR horizontal rows of the gray area. W + FS-1 the pixel values in the last horizontal row of the white area are copied to the pixel values in the last SR horizontal rows of the gray area. Next, the H+FS-1 pixel values in the first vertical row of the white area, as well as the SR pixel values of the gray area obtained on the top side and the SR pixel values of the gray area obtained on the bottom side, can be copied to the pixel values in the first SR vertical rows gray area. H+FS-1 pixel values in the last vertical row of the white area and SR pixel values of the gray area obtained on the top side and SR pixel values of the gray area obtained on the bottom side can be copied to the pixel values in the last SR vertical rows of the gray area.

[316] In another example, the H+FS-1 pixel values in the first vertical row of the white area can first be copied to the pixel values in the first SR vertical rows of the gray area. The H+FS-1 pixel values in the last vertical row of the white area are copied to the pixel values in the last SR vertical rows of the gray area. Further, W+FS-1 pixel values in the first horizontal row of the white area, SR pixel values of the gray area obtained on the left side, and SR pixel values of the gray area obtained on the right side can be copied to the pixel values in the first SR horizontal rows of the gray area . W+FS-1 pixel values in the last horizontal row of the white area, SR pixel values of the gray area obtained on the left side, and SR pixel values of the gray area obtained on the right side can be copied to the pixel values in the last SR horizontal rows of the gray area.

[317] Embodiment 32: After obtaining the target motion vector, based on the target motion vector of each sub-block, bi-directional prediction values (i.e., three components Y, U, V, i.e., the prediction value of the third reference block and the value reference block prediction) with an 8-tap interpolation filter and weight them to obtain final prediction values. On FIG. 5, the pixel values of the black region and the white region are obtained from the reference frame, and the pixel values of the gray region are obtained by copying directly from the corresponding area of the reference frame, rather than copying adjacent pixel values. This method is simple and provides better performance, but it increases the amount of access data per reference frame.

[318] Embodiment 33: After obtaining the target motion vector, based on the target motion vector of each sub-block, bi-directional prediction values (i.e., three Y, U, V components, i.e., the predicted value of the third reference block and the predicted the value of the fourth reference block) with a bilinear interpolation filter (rather than an 8-tap interpolation filter) and weight them to obtain final prediction values. In Fig. 5, the pixel values of the black region and the white region are obtained from the reference frame. Since the number of taps is relatively small, there is no need for gray area pixel values.

[319] Embodiment 34: In Embodiments 30 to 33, after obtaining the prediction values in the two directions, the final prediction value is obtained as an arithmetic weighted average (i.e., the prediction values in the two directions have the same weight). Alternatively, after bidirectional prediction values are obtained, the final predicted value is obtained as a weighted average, and the weights of the two prediction values may be different. For example, the weight ratio of two prediction values may be 1:2, 1:3, 2:1, and so on.

[320] For the encoding side, the weight table may include a weight ratio: 1:2, 1:3, 2:1, etc., while the encoding side can determine the value of the cost of each weight ratio and determine the weight ratio with a minimum cost value, whereby the encoding side can obtain the final prediction value as a weighted average based on the ratio of the weights with the minimum cost value.

[321] When the coding side sends the coding bitstream to the decoding side, the coding bitstream contains a weight ratio index value in the weight table. Thus, the decoding side analyzes the index value of the encoding bitstream, obtains a weight ratio corresponding to the index value from the weight table, and obtains a prediction end value as a weighted average based on the weight ratio.

[322] In one example, the weight table may include, among other things, {-2, 3, 4, 5, 10}. The weight "-2" indicates that the final prediction value = (prediction value 1*(-2) + prediction value 2*(8- (-2))), i.e. (-2*-prediction value 1 + 10*-prediction value 2) / 8. The weight "10" indicates that the weight ratio is 10: -2, i.e. final prediction value = (prediction value 1*(10) + predicted value 2*(-2)), i.e. (10*prediction value 1 - 2*prediction value 2) / 8. Weight "3" indicates a weight ratio of 3:5. Weight "5" indicates a weight ratio of 5:3. Weight "4" indicates a weight ratio of 4:4, m .e. weight "4" indicates that the weights are the same.

[323] Embodiment 35: At step 305 and step 405, the first and second target motion vectors of each subblock of the current block are to be stored, with the first and second target motion vectors being used as references for encoding and decoding the subsequent block. For example, the first and second target motion vectors are used for loop filtering of the current frame; the first and second target motion vectors are used for time-domain anchoring of subsequent frames; and/or the first and second target motion vectors are used for anchoring in the spatial region of the current frame.

[324] For example, the first and second target motion vectors of each sub-block of the current block can be used to compensate for the motion of the current block, as well as to anchor subsequent frames in the time domain. In another example, the first and second target motion vectors of each sub-block of the current block can be used to compensate for the motion of the current block, to contour filter the current block, and to time-domain anchor subsequent frames.

[325] In another example, the first and second target motion vectors of each sub-block of the current block can be used to compensate for the motion of the current block, to loop filter the current block, to anchor in the time domain of subsequent frames, and also for anchor in the spatial domain of the current frame, as will be disclosed below.

[326] The first and second target motion vectors of each sub-block of the current block can be used to anchor in the spatial domain of blocks within certain LECs (Largest Coding Units (HEC)) in the spatial domain. Since encoding is performed in a top-to-bottom, left-to-right sequence, the current block's motion vector can be linked to other blocks within the current NEK, as well as blocks within a subsequent adjacent NEK. Given the large amount of computation required to obtain the target motion vector, if subsequent blocks are tied to the target motion vector of the current block, the wait time will be long. In order to avoid time delay due to too long waiting, it is allowed to bind only a small number of adjoining blocks in the spatial region to the target motion vector of the current block, and other blocks are bound to the source motion vector of the current block. In FIG. 8, the indicated small number of blocks includes sub-blocks located in the largest coding units located below and below to the right of the bottom side of the current HEK, while sub-blocks located in the largest coding units located to the right and lower left may not be associated. to the target motion vector of the current block.

[327] Embodiment 36: A particular example of a motion vector correction process is disclosed below. The specific motion vector correction steps may be as follows, where "copy" means to obtain without interpolation, i.e. if the VD is an integer pixel offset, then it can be directly copied from the reference frame, otherwise interpolation is necessary.

[328] In step e1, if the merge or skip mode is applied to the current block, post-processing is performed.

[329] In step e2, reference pixel values are prepared (assuming the width of the current block is W and the height is H).

[390] Integer pixel block for step e3 is prepared: based on the original motion vector (original motion vector list0 is Org_MV0, original motion vector list1 is Org_MV1), integer pixel blocks of three components with two area blocks (W+FS-1) (H+FS -1) are copied at the corresponding position of the corresponding reference frames.

[331] Whole pixel block is prepared for step e4: based on whole pixel blocks with area (W+FS-1)*(H+FS-1), whole pixel blocks of three components with area (W+FS-1) (H+FS-1) increase up, down, left and right by SR horizontal rows/vertical rows and get, after enlargement, whole pixel blocks of three components of area (W+FS-1+2*SR) * (H+FS-1+2*SR) denoted 5 as Pred_Inter0, Pred_Inter1.

[332] First, motion compensation is performed based on motion information in two different directions. For example, for the luminance component (since only the luminance component is used to calculate the cost value for simplicity in the subsequent search process), the original reference pixel prediction values (denoted Pred_Bilinear0, Pred_Bilinear1) of size (W +2 SR)*((H +2 SR)) can be obtained by bilinear interpolation based on two integer pixel reference blocks of area (W+FS-1)*(H+FS-1). FS is the number of filter taps, which is 8 by default, and SR is the search range, i.e. the maximum interpolation of the horizontal/vertical component of the target motion vector and the source motion vector, defaulting to 2. Pred_Bilinear0/Pred_Bilinea1 is applied in step e3.

[333] In step e3, for each sub-block dx*dy (an entire block of 16*16 or less) of the current block, corresponding target motion vectors are obtained (two target motion vectors are designated Refined_MV0 and Refined_MV1, respectively).

[334] At step e31, SR iterations are performed to obtain the optimal integer pixel offset of the integer VD point, denoted by IntegerDeltaMV, with IntegerDeltaMV set to the initial value (0, 0), and subsequent processing is performed for each iteration:

[335] at step e311 set the value (0, 0) for deltaMV. When the first iteration is performed, two prediction value blocks (ie, the closest to the center block W*H Pred_Bilinear0/1) are obtained by copying based on the original motion vectors in the reference pixel Pred_Bilinear0/1; and get the initial cost value, i.e. A CAP after a vertical downsampling operation on bi-directional prediction value blocks by a factor of 2 based on said two prediction value blocks.

[336] If the initial cost value is less than 4*dx*dy/2, where dx and dy are the width and height of the current subblock, the subsequent search process is directly skipped, step e32 is performed, and notZeroCost is set to "false".

[337] In step e312, as shown in FIG. 6, five offset VDs (all referred to as MVOffset) are sequentially obtained in the sequence {Mv(0,1), Mv(0,-1), Mv(1,0), Mv(-1.0), Mv(right, down)}, with the above starting point taken as the center, and a process of calculating and comparing cost values of said five offset PDs is performed.

[338] For example, based on the determined MVOffset, two blocks of prediction value are obtained at the reference pixel Pred_Bilinear0/1 (e.g., a W*H block with a center offset of MVOffset in Pred_Bilinear0 and a W*H block with a center offset of -MVOffset (opposite of list0) in Pred_Bilinear1) based on the given MVOffset, the CAP of the two blocks is downsampled, and the prediction values are calculated as the cost value of the MVOffset. Next, an MVOffset is reserved (stored in deltaMV) with a minimum cost value, which is used as the new center offset point of the next iteration.

[339] The default value of Mv(right, down) is (-1, -1), while if the cost value of Mv(1,0) is less than the value of Mv(-1, 0), then the value of "right » is 1; if the cost value of Mv(0, 1) is less than the value of Mv(0, -1), then the down value is 1.

[340] IntegerDeltaMV is updated based on the value of deltaMV: IntegerDeltaMV:= IntegerDeltaMV + deltaMV.

[341] In step e313, after iteration, if the optimal VP is still the original VP or the minimum cost value is 0, then the next iterative search process is not performed, step e32 is performed, and notZeroCost is set to false. Otherwise, if the number of iterations has reached SR, step e32 is performed, and if the number of iterations has not reached SR, the optimal VD is taken as the center and the next iterative search process is performed, that is, return to step e311.

[342] In step e32, take the optimal point of the whole-pixel VD in step e31 as the center, obtain the optimal sub-pixel offset VD, which is designated SPMV, set the initial value (0, 0) for SPMV, and then perform the following procedure:

[343] In step e321, the following procedure is performed only if notZeroCost is not "false" and deltaMV is (0, 0), otherwise the original motion vector is corrected directly using IntegerDeltaMV.

[344] In step e322, E(x, y) is presented as a cost value corresponding to the offset PD of the optimal PD point (x, y) obtained in step e31 (cost value calculated in step e31). Based on E (x, y) of the center, top, bottom, left and right points, you can get the offset (x ₀ , y ₀ ) of the point with the smallest E(x, y) as follows: x ₀ =N*(E(-1 ,0)-E(1,0)) / (E(-1,0) + E(1,0)-2* E(0,0)), y ₀ =N*(E(0,-1 )-E(0,1)) / (E(0,-1) + E(0,1)-2* E(0,0)).

[345] In one example, if the motion vector pixel precision is 1/2, 1/4, 1/8, and 1/16, then N=1, 2, 4, and 8, respectively. Further, (x ₀ , y ₀ ) is assigned to deltaMv, SPMV = deltaMv/2N, and if the pixel precision of the current motion vector is 1/16, then SPMV may be (x ₀ /16, y ₀ /16).

[346] If E(-1, 0) = E(0, 0), shift half a pixel horizontally to the left (deltaMv[0] = -N).

[347] If E(1, 0) = E(0, 0), shift half a pixel horizontally to the right (deltaMv[0] = N).

[348] If E(0, -1) = E(0, 0), perform a shift of half a pixel vertically up (deltaMv [1] = -N).

[349] If E(0, -1) = E(0, 0), perform a shift of half a pixel vertically down (deltaMv[1] = N).

[350] In step e33, an optimal offset VD is obtained based on IntegrarteltaMV with integer pixel offset in step e31 and SPMV with subpixel offset in step e32, which is designated BestMVoffset. BestMVoffset=IntegerDeltaMV+SPMV. Based on BestMVoffset, you can get target motion vectors in two directions: Refined_MV0= Org_MV0+BestMVoffset; Refined_MV1=Org_MV1-BestMVoffset.

[351] In step e4, bi-directional prediction values are obtained with an 8-tap interpolation filter based on the target motion vector of each subblock and weighted to obtain a final prediction value (three components). For example, based on the target motion vectors Refined_MV0 and Refined_MV1 of each sub-block, in Pred_Inter0/1 obtained in step e2, a corresponding prediction block is obtained by interpolation (the motion vectors may be sub-pixel, and therefore interpolation is necessary to obtain the corresponding block of pixels).

[352] At step e5, the target motion vector is used to compensate for the motion of the current block and anchor in the time domain of the subsequent frame.

[353] Embodiment 37: The above embodiments can be implemented individually or in any combination. For example, embodiments 13, 15, 24, and 29 may be implemented individually. Embodiment 13 and Embodiment 15 are implemented in combination, Embodiment 13 and Embodiment 24 are implemented in combination, Embodiment 13 and Embodiment 29 are implemented in combination, Embodiment 15 and Embodiment 24 are implemented in combination, Embodiment 15 and Embodiment 29 are implemented in combination, Embodiment 24 and Embodiment 29 are implemented in combination, Embodiment 13, Embodiment 15 and Embodiment 24 are implemented in combination, Embodiment 13, Embodiment 15 and Embodiment 29 are implemented in combination, Embodiment 15, Embodiment 24 and Embodiment 29 are implemented in combination, Embodiment 13, Embodiment 15, Embodiment 24 and Embodiment 29 are implemented in combination, and so on. Of course, the above are only some non-limiting examples. All embodiments of the present application can be implemented individually or in combination.

[354] Embodiment 38:

[355] Based on the same idea that underlies the method, one embodiment of the present application also relates to encoding and decoding equipment with the possibility of application on the encoding side or the decoding side, as disclosed in Fig. 9, which is a block diagram of the equipment and the equipment includes:

[356] A determination module 91, configured, if the characteristic information of the current block meets a specific condition, to determine the first reference block corresponding to the current block based on the first original motion vector of the current block, and to determine the second reference block corresponding to the current block based on the second the initial motion vector of the current block; a processing unit 92 configured to correct the first and second source motion vectors according to the first pixel value of the first reference block and the second pixel value of the second reference block to obtain a first target motion vector corresponding to the first source motion vector and a second target motion vector corresponding to the second source motion vector; and an encoding and decoding module 93, configured to encode or decode the current block based on the first and second target motion vectors.

[357] The characteristic information includes at least one of the following: a motion information prediction mode corresponding to the current block; a motion information attribute corresponding to the current block; information about the size of the current block.

[358] If the characteristic information is the motion information prediction mode corresponding to the current block, the determination module 91 is also configured to: determine that the motion information prediction mode corresponding to the current block meets a specific condition if the motion information prediction mode , corresponding to the current block, represents the standard merge mode; or determining that the motion information prediction mode corresponding to the current block meets a special condition if the motion information prediction mode corresponding to the current block is a merge mode to generate a new prediction value by combining the inter prediction value and the intra prediction value.

[359] If the characteristic information is a motion information attribute corresponding to the current block, the determining module 91 is also configured to: determine that the motion information attribute corresponding to the current block meets a special condition if the motion information attribute corresponding to the current block is that the movement information of the current block includes movement information in two different directions; or determining that the motion information attribute corresponding to the current block meets a specific condition if the motion information attribute corresponding to the current block is that the motion information of the current block includes motion information in two different directions, and the distances between the current frame and each of the two reference frames corresponding to the motion information in two different directions are the same; or determining that the motion information attribute corresponding to the current block meets a specific condition if the motion information attribute corresponding to the current block is that motion information of surrounding blocks is reused for the current block; or determining that the motion information attribute corresponding to the current block meets a specific condition if the motion information attribute corresponding to the current block is that the motion information of each sub-block of the current block is the same.

[360] If the characteristic information is current block size information, and the current block size information includes a width value and a height value of the current block, the determination module 91 is also configured to: determine that the current block size information corresponds to a special condition if the width value of the current block is within the first interval [first threshold, second threshold]; or determining that the size information of the current block meets a special condition if the height value of the current block is within the second interval [third threshold, fourth threshold]; or determining that the size information of the current block meets a specific condition if the area obtained based on the width value and the height value of the current block is within a third interval [fifth threshold, sixth threshold]; or determining that the size information of the current block meets a special condition, if the width value is within the first interval [first threshold, second threshold], the height value is within the second interval [third threshold, fourth threshold], and the area is within third interval [fifth threshold, sixth threshold].

[361] The determining module 91 is configured to determine the first reference block corresponding to the current block based on the first original motion vector of the current block and determine the second reference block corresponding to the current block based on the second original motion vector of the current block, which, in particular, means the ability to: determine the first reference block corresponding to the current block from the first reference frame based on the first source motion vector of the current block; and determining a second reference block corresponding to the current block from the second reference frame based on the second source motion vector of the current block; wherein the first pixel value of each pixel point in the first reference block is obtained by interpolating pixel values of adjacent pixel points in the first reference block, or obtained by copying pixel values of adjacent pixel points in the first reference block; the second pixel value of each pixel point in the second reference block is obtained by interpolating the pixel values of adjacent pixel points in the second reference block or by copying the pixel values of adjacent pixel points in the second reference block.

[362] The processing module 92 is configured to correct the first and second source motion vectors according to the first pixel value of the first reference block and the second pixel value of the second reference block to obtain a first target motion vector corresponding to the first source motion vector and a second target motion vector corresponding to to the second initial motion vector, which, in particular, means the possibility of:

[363] correcting the first and second source motion vectors according to the first and second pixel values to obtain the first and second sub-block target motion vectors for each sub-block of the current block, if the current block contains at least one sub-block.

[364] For example, determining the first integer pixel motion vector correction value and the second integer pixel motion vector correction value of the subblock and/or the first subpixel motion vector correction value and the second subpixel motion vector correction value of the subblock based on the first and second pixel values; correcting the first source motion vector according to the first integer motion vector correction value and/or the first subpixel motion vector correction value to obtain the first target motion vector of the subblock; and adjusting the second source motion vector according to the second integer motion vector correction value and/or the second subpixel motion vector correction value to obtain a second target motion vector of the subblock.

[365] The processing unit 92 is configured to determine the first and second integer pixel sub-block motion vector correction values, the first and second sub-pixel sub-block motion vector correction values based on the first and second pixel values, which specifically means being able to:

[366] determining the first original motion vector or the second original motion vector as the central motion vector;

[367] determining the edge motion vector corresponding to the central motion vector;

[368] obtaining a first cost value corresponding to the center motion vector and a second cost value corresponding to the edge motion vector based on the first and second pixel values;

[369] selecting a motion vector from the center motion vector and the edge motion vector as the optimal motion vector based on the first cost value and the second cost value;

[370] evaluating whether an end condition is met; if not, determining the optimal motion vector as the center motion vector and re-determining the edge motion vector corresponding to the center motion vector;

[371] if so, determining first and second integer sub-unit motion vector correction values based on the optimal motion vector; and determining first and second sub-pixel sub-block motion vector correction values based on the optimal motion vector.

[372] The processing module 92 is configured to determine an edge motion vector corresponding to the center motion vector, which specifically means the ability to:

[373] shifting the central motion vector (x, y) in different directions by S and successively obtaining the edge vector (x, y+S) of the motion, the edge vector (x, y-S) of the motion, the edge vector (x+ S, y) of the motion, edge vector (x-S, y) of movement and edge vector (x+ to the right, y+ down) of movement in different directions; or

[374] shifting the central motion vector (x, y) in different directions by S and successively obtaining the edge vector (x, y-S) motion, edge vector (x, y+S) motion, edge vector (x-S, y) motion, edge vector (x+S, y) of movement and edge vector (x+ to the right, y+ down) of movement in different directions;

[375] wherein the default value of the edge vector (x + right, y + down) of motion is (x-S, y-S);

[376] if the cost value of the edge vector (x+S, y) of motion is less than the cost value of the edge vector (x-S, y) of motion, then the value of "right" is S; if the cost value of the edge motion vector (x, y+S) is less than that of the edge motion vector (x, y-S), then the "down" value is S; or, if the cost value of the edge vector (x+S, y) of motion is less than or equal to the cost value of the edge vector (x-S, y) of motion, then the value of "right" is S; if the edge motion vector cost value (x, y + S) is less than or equal to the edge motion vector cost value (x, y-S) then the down value is S.

[377] The processing module 92 is configured to obtain a first cost value corresponding to the center motion vector and a second cost value corresponding to the edge motion vector based on the first and second pixel values, which specifically means being able to:

[378] deriving a first cost value corresponding to the center motion vector and a second cost value corresponding to the edge motion vector based on the first and second non-downsampling pixel values; or

[379] performing a downsampling operation on the first and second pixel values; and deriving a first cost value corresponding to the center motion vector and a second cost value corresponding to the edge motion vector based on the first and second downsampled pixel values; or

[380] performing a shift and downsampling operation on the first pixel value, and performing a shift and downsampling operation on the second pixel value; and deriving a first cost value corresponding to the center motion vector and a second cost value corresponding to the edge motion vector based on the first and second pixel values after the shift and downsampling.

[381] The processing unit 92 is configured to determine the first and second integer pixel sub-block motion vector correction values, the first and second sub-pixel sub-block motion vector correction values based on the first and second pixel values, which specifically means being able to:

[382] taking the first source motion vector or the second source motion vector as the center, selecting some or all of the motion vectors from the surrounding motion vectors, and determining the selected motion vectors as potential motion vectors;

[383] obtaining a third cost value corresponding to the first source motion vector or the second source motion vector and a fourth cost value corresponding to each of the potential motion vectors based on the first and second pixel values;

[384] selecting a motion vector from the first original motion vector or the second original motion vector and the potential motion vector as the optimal motion vector based on the third cost value and the fourth cost value;

[385] determining first and second integer pixel motion vector correction values of the subblock based on the optimal motion vector; and determining first and second sub-pixel sub-block motion vector correction values based on the optimal motion vector.

[386] The processing unit 92 is configured to determine the first and second integer pixel motion vector correction values of the subblock based on the optimal motion vector, which specifically means: determining the first integer pixel motion vector correction value of the subblock based on the optimal motion vector and the first source vector motion and determining a second integer pixel motion vector correction value of the subblock based on the first integer pixel motion vector correction value.

[387] The processing unit 92 is configured to determine the first and second sub-pixel sub-block motion vector correction values based on the optimal motion vector, which specifically means: determining the first sub-pixel sub-block motion vector correction value based on the cost value corresponding to the optimal motion vector , and a cost value corresponding to an edge motion vector corresponding to the optimal motion vector, and determining a second sub-pixel motion vector correction value of the sub-block based on the first sub-pixel motion vector correction value.

[388] Encoding and decoding module 93 is configured to encode or decode the current block based on the first and second target motion vectors, which specifically means: if the current block contains at least one subblock, definitions for each subblock of the current block , a third reference block corresponding to the subblock based on the first target motion vector of the subblock, and determining a fourth reference block corresponding to the subblock based on the second target motion vector of the subblock; weighting a third pixel value of the third reference block and a fourth pixel value of the fourth reference block to obtain a sub-block prediction value; and determining a prediction value of the current block based on the prediction value of each sub-block.

[389] The encoding and decoding module 93 is configured to determine the third reference block corresponding to the subblock based on the first target motion vector of the subblock, and to determine the fourth reference block corresponding to the subblock based on the second target motion vector of the subblock, which specifically means being able to: determine a fifth reference block corresponding to the sub-block from the first reference frame based on the first target motion vector of the sub-block and interpolate pixel values in the fifth reference block to obtain a third reference block; determining a sixth reference block corresponding to the subblock from the second reference frame based on the second target motion vector of the subblock, and interpolating pixel values in the sixth reference block to obtain a fourth reference block.

[390] The encoding and decoding module 93 is configured to determine the third reference block corresponding to the subblock based on the first target motion vector of the subblock, and to determine the fourth reference block corresponding to the subblock based on the second target motion vector of the subblock, which specifically means that : determining a seventh reference block corresponding to the sub-block from the first reference frame based on the first target motion vector of the sub-block, constructing an eighth reference block based on the pixel values in the seventh reference block, and interpolating the pixel values in the eighth reference block to obtain a third reference block; determining a ninth reference block corresponding to the subblock from the second reference frame based on the second target motion vector of the subblock, constructing a tenth reference block based on the pixel values in the ninth reference block, and interpolating the pixel values in the tenth reference block to obtain the fourth reference block.

[391] The equipment also includes: a storage module for storing the first and second target motion vectors for the current block; wherein the first and second target motion vectors are used for loop filtering of the current frame, the first and second target motion vectors are used for anchoring in the time domain of subsequent frames, and/or the first and second target motion vectors are used for anchoring in the spatial domain of the current frame.

[392] Embodiment 39:

[393] The device on the decoding side of the embodiment of the present application at the hardware level, the hardware structure diagram of which can be seen, in particular, in Fig.10. The device includes: a processor 101 and a computer-readable storage medium 102, the computer-readable storage medium 102 having computer-executable instructions stored thereon, which the processor 101 can execute; the processor 101 is configured to execute computer-executable instructions for implementing the methods disclosed in the above examples of the present application. For example, a processor is used to execute machine-executable instructions to implement steps in which:

[394] if the characteristic information of the current block meets the specific condition, determining the first reference block corresponding to the current block based on the first original motion vector of the current block, and determining the second reference block corresponding to the current block based on the second original motion vector of the current block; adjusting the first and second source motion vectors according to the first pixel value of the first reference block and the second pixel value of the second reference block to obtain a first target motion vector corresponding to the first source motion vector and a second target motion vector corresponding to the second source motion vector; and decoding the current block based on the first and second target motion vectors.

[395] The device on the side of encoding according to an embodiment of the present application at the hardware level, the hardware structure diagram of which can be seen, in particular, in Fig.11. The device includes: a processor 111 and a computer-readable storage medium 112, the computer-readable storage medium 112 having computer-executable instructions stored thereon, which can be executed by the processor 111; the processor 111 is configured to execute computer-executable instructions for implementing the methods disclosed in the above examples of the present application. For example, the processor 111 is used to execute computer-executable instructions to implement the steps of: if the characteristic information of the current block meets a specific condition, determine the first reference block corresponding to the current block based on the first initial motion vector of the current block, and determine the second reference block corresponding to the current block, based on the second source motion vector of the current block; adjusting the first and second source motion vectors according to the first pixel value of the first reference block and the second pixel value of the second reference block to obtain a first target motion vector corresponding to the first source motion vector and a second target motion vector corresponding to the second source motion vector; and encoding the current block based on the first and second target motion vectors.

[396] Based on the same idea that underlies the method, an embodiment of the present application also relates to a computer-readable storage medium, and the computer-readable storage medium contains a plurality of computer instructions stored thereon, while, when the computer instructions are executed by the processor, can be implemented the encoding and decoding methods disclosed in the above examples of the present application. A computer-readable storage medium can be any electronic, magnetic, optical, or other physical storage device capable of containing or storing information, such as executable instructions, data, and the like. For example, a computer-readable storage medium can be: any type (for example, a compact disc, a digital versatile disk (DVD-ROM, English Digital Versatile Disk), etc.), or a similar storage medium, or a combination thereof.

[397] The systems, equipment, modules, or blocks discussed in the embodiments disclosed above may, in particular, be implemented by a computer chip or object, or implemented by a product with certain functions. A typical implementation device is a computer, which may be in the form of a personal computer, laptop, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email messaging device, game console, tablet computer, wearable device, or their combinations.

[398] To simplify the description, the above devices are disclosed separately in various blocks in accordance with their functions. Of course, in the implementation of the present application, the functions of each block can be implemented in one or more software and/or hardware.

[399] Those skilled in the art will appreciate that embodiments of the present application may be disclosed in connection with methods, systems, or computer program products. Accordingly, the present application may be in the form of an all-hardware embodiment, an all-software embodiment, or a combination of software and hardware aspects. In addition, embodiments of the present application may be a computer program product implemented on one or more media suitable for use in a computer (including, but not limited to, a disk drive, compact disk read only memory). -Only Memory (CD-ROM)), an optical drive, etc.), with a program executed in it, suitable for use in a computer.

[400] Embodiments of methods, devices (systems) and computer software products are disclosed in this application by examples illustrated in the sequence diagrams and/or block diagrams. It should be understood that each sequence diagram and/or block of sequence diagrams and/or block diagrams and combination of sequence diagrams and/or blocks of sequence diagrams and/or block diagrams may be implemented by computer program instructions. Computer program instructions may be entered into the processor of a general purpose computer, a special purpose computer, an embedded processor, or other programmable data processing device, resulting in a machine by which instructions, executable by the computer processor or other programmable data processing device, form equipment. to implement the functions indicated in the sequence or sequences in the sequence diagrams and/or in the block or blocks of the block diagrams.

[401] In addition, computer program instructions may also be stored in a computer-readable storage device that can cause a computer or other data processing firmware to function in a certain way, in connection with which the instructions stored in this computer-readable storage device form an article, including a control device that implements the function indicated in the sequence or sequences in the sequence diagrams and/or in the block or blocks of the block diagrams.

[402] Computer program instructions may also be downloaded to a computer or other programmable processing device to cause the computer or other programmable device to execute a series of operating steps to form a computerized process in which the instructions executable on the computer or other programmable device are referred to as steps. implementation of the functions indicated in the sequence or sequences in the sequence diagrams and/or in the block or blocks of the block diagrams.

[403] The above discloses only one of the embodiments of the present application, not intended to limit the present application. Specialists in the field of technology to which this application relates, may come to mind a variety of modifications and changes. Any modification, equivalent replacement, improvement, or the like, without departing from the spirit and principle of the present application, is within the scope of the claims of the present application.

Claims (68)

1. A decoding method, which includes the steps of: when the current block allows the motion vector refinement mode, determining the sub-block prediction values of the current block, and determining the prediction value of the current block based on the sub-block prediction values of the current block, wherein for each sub-block of the current block, determining the sub-block prediction value includes the following: receiving information about the movement of the current block, containing the first source motion vector and the first reference frame, the second source motion vector and the second reference frame; determining a first reference block corresponding to the subblock based on the first source motion vector of the current block, and determining a second reference block corresponding to the subblock based on the second source motion vector of the current block; deriving the optimal motion vector based on the pixel values of the first reference block and the pixel values of the second reference block, wherein obtaining the optimal motion vector includes the following: taking the original motion vector as the center, selecting motion vectors from motion vectors containing the original motion vector and motion vectors surrounding the original motion vector, and determining the selected motion vectors as candidate motion vectors; wherein the original motion vector is a first original motion vector or a second original motion vector; obtaining a cost value corresponding to the original motion vector and cost values corresponding to the potential motion vectors based on the pixel values of the first reference block and the pixel values of the second reference block; selecting a motion vector from the original motion vector and the potential motion vectors as the optimal motion vector based on the cost value corresponding to the original motion vector and the cost values corresponding to the potential motion vectors; adjusting the first source motion vector and the second source motion vector according to the optimal motion vector to obtain a first target motion vector corresponding to the first source motion vector and a second target motion vector corresponding to the second source motion vector; determining a subblock prediction value based on the first target motion vector and the second target motion vector; wherein determining the sub-block prediction value based on the first target motion vector and the second target motion vector includes: determining a third reference block corresponding to the sub-block based on the first target motion vector, and determining a fourth reference block corresponding to the sub-block based on the second target motion vector movement; weighing the pixel values of the third reference block and the pixel values of the fourth reference block to obtain a sub-block prediction value. 2. The method according to claim 1, in which: the step of taking the original motion vector as the center, selecting the motion vectors from the motion vectors containing the original motion vector and motion vectors surrounding the original motion vector, and determining the selected motion vectors as potential motion vectors, includes: receiving the original motion vector motion as the center, search for 25 motion vectors containing the original motion vector in the search range of 2, and determine 25 motion vectors as candidate motion vectors; wherein the search sequence of 25 motion vectors is: { Mv(-2, -2), Mv(-1, -2), Mv(0, -2), Mv(1, -2), Mv(2, - 2), Mv(-2, -1), Mv(-1, -1), Mv(0, -1), Mv(1, -1), Mv(2, -1), Mv(-2, 0), Mv(-1, 0), Mv(0, 0), Mv(1, 0), Mv(2, 0), Mv(-2, 1), Mv(-1, 1), Mv( 0, 1), Mv(1, 1), Mv(2, 1), Mv(-2, 2), Mv(-1, 2), Mv(0, 2), Mv(1, 2), Mv (2, 2) }. 3. The method according to claim 1, in which: a step of obtaining a cost value corresponding to the original motion vector and cost values corresponding to the potential motion vectors based on the pixel values of the first reference block and the pixel values of the second reference block, includes the following: obtaining cost values corresponding to potential motion vectors includes: determining first reference subblocks corresponding to potential motion vectors based on the first reference block, and determining second reference subblocks corresponding to potential motion vectors based on the second reference block; calculate the sum of the absolute differences of the sampled values of the pixel dots in the first reference sub-blocks and the second reference sub-blocks by using a 2-fold vertical downsampling operation; determining cost values corresponding to potential motion vectors based on the obtained sum of absolute differences; obtaining a cost value corresponding to the original motion vector includes: determining a first reference subblock corresponding to the original motion vector based on the first reference block, and determining a second reference subblock corresponding to the original motion vector based on the second reference block; calculate the sum of the absolute differences of the sampled values of the pixel dots in the first reference subblock and the second reference subblock by using a 2-fold vertical downsampling operation; determine the value of the cost corresponding to the original motion vector, based on the obtained sum of absolute differences. 4. The method according to claim 1, in which: the step of adjusting the first source motion vector and the second source motion vector according to the optimal motion vector to obtain the first target motion vector corresponding to the first source motion vector and the second target motion vector corresponding to the second source motion vector, includes: determining a first integer pixel motion vector correction value and a second integer pixel motion vector correction value based on the optimal motion vector; correcting the first source motion vector according to the first integer pixel motion vector correction value to obtain a first target motion vector corresponding to the first source motion vector; and correcting the second source motion vector according to the second integer pixel motion vector correction value to obtain a second target motion vector corresponding to the second source motion vector; or a step of determining the first integer pixel motion vector correction value and the second integer pixel motion vector correction value based on the optimal motion vector, includes: determining a first integer pixel motion vector correction value based on the optimal motion vector, the first integer pixel motion vector correction value being equal to the difference between the optimal motion vector and the original motion vector; determining a second integer pixel motion vector correction value based on the first integer pixel motion vector correction value; wherein the second integer pixel motion vector correction value and the first integer pixel motion vector correction value are opposite to each other; or a step of adjusting the first source motion vector according to the first integer pixel motion vector correction value to obtain a first target motion vector corresponding to the first source motion vector; and adjusting the second source motion vector according to the second integer motion vector correction value to obtain a second target motion vector corresponding to the second source motion vector, including: the first target motion vector is equal to the sum of the first source motion vector and the first integer pixel motion vector correction value; the second target motion vector is equal to the sum of the second source motion vector and the second integer pixel motion vector correction value. 5. The method according to claim 1, in which the step of adjusting the first source motion vector and the second source motion vector according to the optimal motion vector to obtain the first target motion vector corresponding to the first source motion vector and the second target motion vector corresponding to the second source motion vector, includes: determining a first integer pixel motion vector correction value, a first subpixel motion vector correction value, a second integer pixel motion vector correction value, and a second subpixel motion vector correction value based on the optimal motion vector; correcting the first source motion vector according to the first integer motion vector correction value and the first subpixel motion vector correction value to obtain the first target motion vector of the subblock; and correcting the second source motion vector according to the second integer motion vector correction value and the second subpixel motion vector correction value to obtain a second target motion vector of the subblock; or a step of determining the first integer-pixel motion vector correction value and the first sub-pixel motion vector correction value based on the optimal motion vector, includes: determining a first integer pixel motion vector correction value based on the optimal motion vector, the first integer pixel motion vector correction value being equal to the difference between the optimal motion vector and the original motion vector; determining a first sub-pixel motion vector correction value based on a cost value corresponding to the optimal motion vector and cost values corresponding to edge motion vectors corresponding to the optimal motion vector; or the step of determining the first sub-pixel motion vector correction value based on the cost value corresponding to the optimal motion vector and the cost values corresponding to edge motion vectors corresponding to the optimal motion vector, includes: successively determine the value of the cost of 5 integer motion vectors, for which the optimal motion vector is taken as the center; wherein the 5 integer motion vectors are 5 edge motion vectors obtained by shifting horizontally left, horizontally right, vertically up, and vertically down, respectively, when the optimal motion vector is taken as the center. 6. The method of claim 1, wherein the step of adjusting the first source motion vector and the second source motion vector according to the optimal motion vector to obtain the first target motion vector corresponding to the first source motion vector and the second target motion vector corresponding to the second source motion vector, includes: determining a first integer pixel motion vector correction value and a first subpixel motion vector correction value based on the optimal motion vector; obtaining an optimal offset motion vector according to the first integer motion vector correction value and the first subpixel motion vector correction value; adjusting the first source motion vector based on the optimal offset motion vector to obtain a first target motion vector; adjusting the second source motion vector based on a value opposite to the optimal offset motion vector to obtain a second target motion vector; or a step of determining the first integer-pixel motion vector correction value and the first sub-pixel motion vector correction value based on the optimal motion vector, includes: determining a first integer pixel motion vector correction value based on the optimal motion vector, the first integer pixel motion vector correction value being equal to the difference between the optimal motion vector and the original motion vector; determining a first sub-pixel motion vector correction value based on a cost value corresponding to the optimal motion vector and cost values corresponding to edge motion vectors corresponding to the optimal motion vector; or the step of determining the first sub-pixel motion vector correction value based on the cost value corresponding to the optimal motion vector and the cost values corresponding to edge motion vectors corresponding to the optimal motion vector, includes: successively determine the value of the cost of 5 integer motion vectors, for which the optimal motion vector is taken as the center; wherein the 5 integer motion vectors are 5 edge motion vectors obtained by shifting horizontally left, horizontally right, vertically up, and vertically down, respectively, when the optimal motion vector is taken as the center. 7. The method of claim 5, wherein the first target motion vector is equal to the sum of the first source motion vector, the first integer pixel motion vector correction value, and the first subpixel motion vector correction value; the second target motion vector is equal to the sum of the second source motion vector, the second integer-pixel motion vector correction value, and the second sub-pixel motion vector correction value; wherein the second integer pixel motion vector correction value and the first integer pixel motion vector correction value are opposite to each other; the second sub-pixel motion vector correction value and the first sub-pixel motion vector correction value are opposite to each other. 8. The method of claim 1, wherein the optimal motion vector is a lower cost motion vector selected from the original motion vector and potential motion vectors. 9. The method of claim 1, wherein when the current block contains one subblock, said subblock is the current block. 10. An encoding method, which includes the steps of: when the current block allows the motion vector refinement mode, determining the sub-block prediction values of the current block, and determining the prediction value of the current block based on the sub-block prediction values of the current block, wherein for each sub-block of the current block, determining the sub-block prediction value includes the following: receiving information about the movement of the current block, containing the first source motion vector and the first reference frame, the second source motion vector and the second reference frame; determining a first reference block corresponding to the subblock based on the first source motion vector of the current block, and determining a second reference block corresponding to the subblock based on the second source motion vector of the current block; deriving the optimal motion vector based on the pixel values of the first reference block and the pixel values of the second reference block, wherein obtaining the optimal motion vector includes the following: taking the original motion vector as the center, selecting motion vectors from motion vectors containing the original motion vector and motion vectors surrounding the original motion vector, and determining the selected motion vectors as candidate motion vectors; wherein the original motion vector is a first original motion vector or a second original motion vector; obtaining a cost value corresponding to the original motion vector and cost values corresponding to the potential motion vectors based on the pixel values of the first reference block and the pixel values of the second reference block; selecting a motion vector from the original motion vector and the potential motion vectors as the optimal motion vector based on the cost value corresponding to the original motion vector and the cost values corresponding to the potential motion vectors; adjusting the first source motion vector and the second source motion vector according to the optimal motion vector to obtain a first target motion vector corresponding to the first source motion vector and a second target motion vector corresponding to the second source motion vector; determining a subblock prediction value based on the first target motion vector and the second target motion vector; wherein determining the sub-block prediction value based on the first target motion vector and the second target motion vector includes: determining a third reference block corresponding to the sub-block based on the first target motion vector, and determining a fourth reference block corresponding to the sub-block based on the second target motion vector movement; weighing the pixel values of the third reference block and the pixel values of the fourth reference block to obtain a sub-block prediction value. 11. Decoding equipment, comprising: a processor and a computer-readable storage medium, and the computer-readable storage medium contains machine-executable instructions stored thereon, which can be executed by the processor; moreover, the processor is configured to be used to execute machine-executable instructions for implementing the method according to any one of paragraphs. 1-9. 12. Encoding equipment, comprising: a processor and a computer-readable storage medium, and the computer-readable storage medium contains machine-executable instructions stored thereon, which can be executed by the processor; moreover, the processor is configured to be used to execute machine-executable instructions to implement the steps, in which: when the current block allows the motion vector refinement mode, determining the sub-block prediction values of the current block, and determining the prediction value of the current block based on the sub-block prediction values of the current block, wherein for each sub-block of the current block, determining the sub-block prediction value includes the following: receiving information about the movement of the current block, containing the first source motion vector and the first reference frame, the second source motion vector and the second reference frame; determining a first reference block corresponding to the subblock based on the first source motion vector of the current block, and determining a second reference block corresponding to the subblock based on the second source motion vector of the current block; deriving the optimal motion vector based on the pixel values of the first reference block and the pixel values of the second reference block, wherein obtaining the optimal motion vector includes the following: taking the original motion vector as the center, selecting motion vectors from motion vectors containing the original motion vector and motion vectors surrounding the original motion vector, and determining the selected motion vectors as candidate motion vectors; wherein the original motion vector is a first original motion vector or a second original motion vector; obtaining a cost value corresponding to the original motion vector and cost values corresponding to the potential motion vectors based on the pixel values of the first reference block and the pixel values of the second reference block; selecting a motion vector from the original motion vector and the potential motion vectors as the optimal motion vector based on the cost value corresponding to the original motion vector and the cost values corresponding to the potential motion vectors; adjusting the first source motion vector and the second source motion vector according to the optimal motion vector to obtain a first target motion vector corresponding to the first source motion vector and a second target motion vector corresponding to the second source motion vector; determining a subblock prediction value based on the first target motion vector and the second target motion vector; wherein determining the sub-block prediction value based on the first target motion vector and the second target motion vector includes: determining a third reference block corresponding to the sub-block based on the first target motion vector, and determining a fourth reference block corresponding to the sub-block based on the second target motion vector movement; weighing the pixel values of the third reference block and the pixel values of the fourth reference block to obtain a sub-block prediction value. 13. Decoding device, configured to execute the method according to any one of paragraphs. 1-9. 14. An encoder configured to execute the steps where: when the current block allows the motion vector refinement mode, determining the sub-block prediction values of the current block, and determining the prediction value of the current block based on the sub-block prediction values of the current block, wherein for each sub-block of the current block, determining the sub-block prediction value includes the following: receiving information about the movement of the current block, containing the first source motion vector and the first reference frame, the second source motion vector and the second reference frame; determining a first reference block corresponding to the subblock based on the first source motion vector of the current block, and determining a second reference block corresponding to the subblock based on the second source motion vector of the current block; deriving the optimal motion vector based on the pixel values of the first reference block and the pixel values of the second reference block, wherein obtaining the optimal motion vector includes the following: taking the original motion vector as the center, selecting motion vectors from motion vectors containing the original motion vector and motion vectors surrounding the original motion vector, and determining the selected motion vectors as candidate motion vectors; wherein the original motion vector is a first original motion vector or a second original motion vector; obtaining a cost value corresponding to the original motion vector and cost values corresponding to the potential motion vectors based on the pixel values of the first reference block and the pixel values of the second reference block; selecting a motion vector from the original motion vector and the potential motion vectors as the optimal motion vector based on the cost value corresponding to the original motion vector and the cost values corresponding to the potential motion vectors; adjusting the first source motion vector and the second source motion vector according to the optimal motion vector to obtain a first target motion vector corresponding to the first source motion vector and a second target motion vector corresponding to the second source motion vector; determining a subblock prediction value based on the first target motion vector and the second target motion vector; wherein determining the sub-block prediction value based on the first target motion vector and the second target motion vector includes: determining a third reference block corresponding to the sub-block based on the first target motion vector, and determining a fourth reference block corresponding to the sub-block based on the second target motion vector movement; weighing the pixel values of the third reference block and the pixel values of the fourth reference block to obtain a sub-block prediction value. 15. A non-volatile computer-readable storage medium containing machine-executable instructions stored thereon that can be executed by a processor, the processor being capable of being used to execute machine-executable instructions for implementing a method according to any one of paragraphs. 1-9.

Images