Radio frequency assembly and electronic device
The present disclosure relates to a motion vector prediction method, a video coding method, and related apparatuses and devices. The motion vector prediction method includes: acquiring a candidate list of a current coding block in a preset prediction mode, where the candidate list includes a plurality of candidate motion vectors; performing a motion search of the plurality of candidate motion vectors in the candidate list to obtain an initial optimal motion vector of the current coding block in the preset prediction mode; performing motion compensation on the initial optimal motion vector by using each of a first preset number of pixel accuracies, to determine a final optimal motion vector of the current coding block, where the preset prediction mode comprises any one of a plurality of prediction modes, and pixel accuracies used in any one of the prediction modes are identical.
1. A motion vector prediction method, comprising:
acquiring a candidate list of a current coding block in a preset prediction mode, wherein the candidate list comprises a plurality of candidate motion vectors;
performing a motion search of the plurality of candidate motion vectors in the candidate list to obtain an initial optimal motion vector of the current coding block in the preset prediction mode;
performing motion compensation on the initial optimal motion vector by using each of a first preset number of pixel accuracies, to determine a final optimal motion vector of the current coding block, wherein the first preset number of pixel accuracies is equal to or greater than 3;
wherein the preset prediction mode comprises any one of a plurality of prediction modes, and pixel accuracies used in any one of the prediction modes are identical;
wherein said performing motion compensation on the initial optimal motion vector by using each of a first preset number of pixel accuracies, to determine a final optimal motion vector of the current coding block comprises:
performing refinement treatment on the initial optimal motion vector by using each of the first preset number of pixel accuracies to obtain a plurality of first offset motion vectors deviating from the initial optimal motion vector;
using each of the first preset number of pixel accuracies as a first target pixel accuracy;
selecting any one of the initial optimal motion vector and the first offset motion vectors whose pixel accuracies are lower than the first target pixel accuracy, as a starting motion vector corresponding to the first target pixel accuracy;
performing refinement treatment on the starting motion vector corresponding to the first target pixel accuracy by using the first target pixel accuracy to obtain a plurality of second offset motion vectors deviating from the starting motion vector;
selecting, from the plurality of second offset motion vectors, a motion vector whose rate-distortion cost satisfies a preset screening condition, as the final optimal motion vector of the current coding block.
2. The method according to claim 1 , wherein the first preset number is an integer and not less than 3;
and/or, the pixel accuracies comprise: a 1/16 pixel accuracy, a ⅛ pixel accuracy, a ¼ pixel accuracy, a ½ pixel accuracy, an integer pixel accuracy, a 2 times pixel accuracy, a 4 times pixel accuracy, a 8 times pixel accuracy, and a 16 times pixel accuracy;
and/or, in any one of the preset prediction modes, a same one of the pixel accuracies corresponds to a same pixel accuracy identifier used to represent the pixel accuracy;
wherein the plurality of preset prediction modes comprise: a conventional AMVP mode, an affine AMVP mode, an AMVP mode based on intra block copy, and a Hash AMVP mode.
3. The method according to claim 2 , wherein the performing a motion search of the plurality of candidate motion vectors in the candidate list to obtain an initial optimal motion vector of the current coding block in the preset prediction mode comprises:
in response to the conventional AMVP mode, performing motion compensation on the candidate motion vectors in the candidate list and calculating Rate Distortion costs (RDCOSTs) based on Sum of Absolute Difference (SAD) thereof;
selecting the candidate motion vector with the smallest RDCOST as a starting motion vector of a TZ search;
performing a TZ search with the starting motion vector to acquire the initial optimal motion vector.
4. The method according to claim 3 , wherein the performing motion compensation on the initial optimal motion vector by using each of a first preset number of pixel accuracies, to determine a final optimal motion vector of the current coding block comprises:
taking the initial optimal motion vector as a starting motion vector, and using each of the first preset number of pixel accuracies to offset the starting motion vector;
selecting the motion vector with the smallest RDCOST from the plurality of offset motion vectors, as a final optimal motion vector.
5. The method according to claim 2 , wherein the performing a motion search of the plurality of candidate motion vectors in the candidate list to obtain an initial optimal motion vector of the current coding block in the preset prediction mode comprises:
in response to the affine AMVP mode, according to a position relationship between the center point and the CP point of each 4*4 sub-block, using a Control Point's MV (CPMV) affine method to obtain the MVs of each sub-block;
after each sub-block acquires the MVs, acquiring a prediction of each sub-block through a motion compensation difference filter, and combining the predictions of all the sub-blocks to derive a pixel value of the entire match block;
making differences between the pixel values of the match blocks and the original pixel value of the current coding block to obtain Rate Distortion costs (RDCOSTs) based on Sum of Absolute Transformed Difference (SATD);
selecting the CPMV with the smallest RDCOST as Control Point's Motion Vector Prediction (CPMVP), using the CPMVP to iteratively update CPMV based on optical flow and thereby acquire the initial optimal motion vector.
6. The method according to claim 5 , wherein the performing motion compensation on the initial optimal motion vector by using each of a first preset number of pixel accuracies, to determine a final optimal motion vector of the current coding block comprises:
subjecting the CPMV subjected to preset offset vectors, and selecting the motion vector with the smallest RDCOST as the final optimal motion vector.
7. The method according to claim 2 , further comprising:
in response to the AMVP mode based on intra block copy, using a method of Hash matching to acquire the initial optimal motion vector.
8. The method according to claim 7 , further comprising:
if the Hash matching fails to find the initial optimal motion vector, using the method of motion search to acquire the initial optimal motion vector.
9. The prediction according to claim 2 , further comprising:
in response to the Hash AMVP mode, using a method of Hash matching to acquire the initial optimal motion vector.
10. The method according to claim 2 , further comprising:
acquiring a candidate list of the current coding block in each of at least two modes of the conventional AMVP mode, the affine AMVP mode, the AMVP mode based on intra block copy, and the Hash AMVP mode;
acquiring the final optimal motion vectors and the corresponding Rate Distortion costs (RDCOSTs) in the at least two modes;
selecting the final optimal motion vector with the smallest RDCOST as the final optimal motion vector of the current coding block.
11. A storage device, having stored thereon program instructions, which can be executed by a processor to implement a motion vector prediction method comprising: acquiring a candidate list of a current coding block in a preset prediction mode, wherein the candidate list comprises a plurality of candidate motion vectors; performing a motion search of the plurality of candidate motion vectors in the candidate list to obtain an initial optimal motion vector of the current coding block in the preset prediction mode; performing motion compensation on the initial optimal motion vector by using each of a first preset number of pixel accuracies, to determine a final optimal motion vector of the current coding block, wherein the first preset number of pixel accuracies is equal to or greater than 3; wherein the preset prediction mode comprises any one of a plurality of prediction modes, and pixel accuracies used in any one of the prediction modes are identical; or to implement a video coding method comprising: constructing a candidate list of a current coding block in a preset prediction mode, wherein the candidate list contains a plurality of candidate motion vectors; acquiring a final optimal motion vector of the current coding block based on the candidate list, wherein the final optimal motion vector is obtained by using a prediction method comprising: acquiring a candidate list of a current coding block in a preset prediction mode, wherein the candidate list comprises a plurality of candidate motion vectors; performing a motion search of the plurality of candidate motion vectors in the candidate list to obtain an initial optimal motion vector of the current coding block in the preset prediction mode; performing motion compensation on the initial optimal motion vector by using each of a first preset number of pixel accuracies, to determine a final optimal motion vector of the current coding block; wherein the preset prediction mode comprises any one of a plurality of prediction modes, and pixel accuracies used in any one of the prediction modes are identical; selecting one of the plurality of candidate motion vectors as a predictive motion vector, and coding the current coding block based on the predictive motion vector and the final optimal motion vector; wherein said performing motion compensation on the initial optimal motion vector by using each of a first preset number of pixel accuracies, to determine a final optimal motion vector of the current coding block comprises: performing refinement treatment on the initial optimal motion vector by using each of the first preset number of pixel accuracies to obtain a plurality of first offset motion vectors deviating from the initial optimal motion vector; using each of the first preset number of pixel accuracies as a first target pixel accuracy; selecting any one of the initial optimal motion vector and the first offset motion vectors whose pixel accuracies are lower than the first target pixel accuracy, as a starting motion vector corresponding to the first target pixel accuracy; performing refinement treatment on the starting motion vector corresponding to the first target pixel accuracy by using the first target pixel accuracy to obtain a plurality of second offset motion vectors deviating from the starting motion vector; selecting, from the plurality of second offset motion vectors, a motion vector whose rate-distortion cost satisfies a preset screening condition, as the final optimal motion vector of the current coding block.