Overlapped block motion compensation for inter prediction
A method for video encoding is provided. The method includes: partitioning a frame of video data into a plurality of blocks; obtaining a current prediction of a current block of the plurality of blocks using one or more motion vectors of the current block; obtaining one or more collocated predictions of the current block using one or more motion vectors of one or more neighbor blocks of the current block in respective one or more blending areas, where each of the one or more blending areas is at least part of the current block and adjustable based at least on the partitioning; and obtaining a combined prediction of the current block by combing the current prediction with the one or more collocated predictions.
1. A method for video encoding, comprising:
partitioning a frame of video data into a plurality of blocks;
obtaining a current prediction of a current block of the plurality of blocks using one or more motion vectors associated with the current block;
obtaining one or more collocated predictions of the current block in one or more blending areas for overlapped block motion compensation (OBMC) by using one or more motion vectors of one or more neighbor blocks of the current block, wherein each of the one or more blending areas is a part of the current block and adjustable based at least on a coding mode of the current block; and
obtaining an OBMC prediction of the current block by performing weighted averaging using a part of the current prediction and the one or more collocated predictions in the one or more blending areas;
wherein in response to determining that the current block is a coding block and the coding mode of the current block is a motion vector angular prediction (MVAP) mode where a motion vector of a sub-block inside the current block is projected from a motion vector of a neighbor block of the current block according to a project direction, the one or more blending areas comprise at least one of left boundary or top boundary of the current block, and each of the one or more collocated predictions of the current block is obtained by using one or more motion vectors of a neighbor block of the current block,
wherein the motion vector of the sub-block inside the current block is calculated by one of following:
p MV[ x][y ]=refMV[−1][ x+y+ 1] for bottom-left direction, or
p MV[ x][y ]=refMV[−1][ y ] for left direction,
wherein pMV denotes the motion vector of the sub-block inside the current block, and ref MV denotes the motion vector of the neighbor blocks of the current block.
2. The method of claim 1 , wherein obtaining the OBMC prediction of the current block comprises:
obtaining the OBMC prediction of the current block by using weights that are adaptively determined based on a height or a width of each of the one or more blending areas.
3. The method of claim 1 , wherein each of the one or more blending areas is adjustable further based on the partitioning.
4. The method of claim 1 , wherein a size of each of the one or more blending areas is determined based on a size of the current block.
5. The method of claim 1 , wherein each of the one or more collocated predictions is generated using one motion vector of a corresponding neighbor block of the one or more neighbor blocks of the current block.
6. An apparatus for encoding video data, comprising:
a memory; and
at least one processor coupled to the memory and configured to perform acts comprising:
partitioning a frame of video data into a plurality of blocks;
obtaining a current prediction of a current block of the plurality of blocks using one or more motion vectors associated with the current block;
obtaining one or more collocated predictions of the current block in one or more blending areas for overlapped block motion compensation (OBMC) by using one or more motion vectors of one or more neighbor blocks of the current block, wherein each of the one or more blending areas is a part of the current block and adjustable based at least on a coding mode of the current block; and
obtaining an OBMC prediction of the current block by performing weighted averaging using a part of the current prediction and the one or more collocated predictions in the one or more blending areas;
wherein in response to determining that the current block is a coding block and the coding mode of the current block is a motion vector angular prediction (MVAP) mode where a motion vector of a sub-block inside the current block is projected from a motion vector of a neighbor block of the current block according to a project direction, the one or more blending areas comprise at least one of left boundary or top boundary of the current block, and each of the one or more collocated predictions of the current block is obtained by using one or more motion vectors of a neighbor block of the current block,
wherein the motion vector of the sub-block inside the current block is calculated by one of following:
p MV[ x][y ]=refMV[−1][ x+y+ 1] for bottom-left direction, or
p MV[ x][y ]=refMV[−1][ y ] for left direction,
wherein pMV denotes the motion vector of the sub-block inside the current block, and ref MV denotes the motion vector of the neighbor blocks of the current block.
7. The apparatus of claim 6 , wherein obtaining the OBMC prediction of the current block comprises:
obtaining the OBMC prediction of the current block by using weights that are adaptively determined based on a height or a width of each of the one or more blending areas.
8. The apparatus of claim 6 , wherein each of the one or more blending areas is adjustable further based on the partitioning.
9. The apparatus of claim 6 , wherein a size of each of the one or more blending areas is determined based on a size of the current block.
10. The apparatus of claim 6 , wherein each of the one or more collocated predictions is generated using one motion vector of a corresponding neighbor block of the one or more neighbor blocks of the current block.
11. A non-transitory computer readable medium storing computer code, which when executed by a processor, causes the processor to perform operations comprising:
partitioning a frame of video data into a plurality of blocks;
obtaining a current prediction of a current block of the plurality of blocks using one or more motion vectors associated with the current block;
obtaining one or more collocated predictions of the current block in one or more blending areas for overlapped block motion compensation (OBMC) by using one or more motion vectors of one or more neighbor blocks of the current block, wherein each of the one or more blending areas is a part of the current block and adjustable based at least on a coding mode of the current block; and
obtaining an OBMC prediction of the current block by performing weighted averaging using a part of the current prediction and the one or more collocated predictions in the one or more blending areas;
wherein in response to determining that the current block is a coding block and the coding mode of the current block is a motion vector angular prediction (MVAP) mode where a motion vector of a sub-block inside the current block is projected from a motion vector of a neighbor block of the current block according to a project direction, the one or more blending areas comprise at least one of left boundary or top boundary of the current block, and each of the one or more collocated predictions of the current block is obtained by using one or more motion vectors of a neighbor block of the current block,
wherein the motion vector of the sub-block inside the current block is calculated by one of following:
p MV[ x][y ]=refMV[−1][ x+y+ 1] for bottom-left direction, or
p MV[ x][y ]=refMV[−1][ y ] for left direction,
wherein pMV denotes the motion vector of the sub-block inside the current block, and ref MV denotes the motion vector of the neighbor blocks of the current block.
12. The non-transitory computer readable medium of claim 11 , wherein obtaining the OBMC prediction of the current block comprises:
obtaining the OBMC prediction of the current block by using weights that are adaptively determined based on a height or a width of each of the one or more blending areas.
13. The non-transitory computer readable medium of claim 11 , wherein each of the one or more blending areas is adjustable further based on the partitioning.
14. The non-transitory computer readable medium of claim 11 , wherein a size of each of the one or more blending areas is determined based on a size of the current block.
15. The non-transitory computer readable medium of claim 11 , wherein each of the one or more collocated predictions is generated using one motion vector of a corresponding neighbor block of the one or more neighbor blocks of the current block.