Image coding method based on affine motion prediction, and device for same
A picture decoding method implemented by a decoding device, according to the present invention, comprises the steps of: acquiring motion prediction information from a bitstream; generating an affine MVP candidate list comprising affine MVP candidates for the current block; deriving CPMVPs for the respective CPs of the current block on the basis of one affine MVP candidate among the affine MVP candidates included in the affine MVP candidate list; deriving CPMVDs for the CPs of the current block on the basis of information on the CPMVDs for the respective CPs included in the acquired motion prediction information; and deriving CPMVs for the CPs of the current block on the basis of the CPMVPs and the CPMVDs.
1. A picture decoding method performed by a decoding apparatus, the method comprising:
obtaining motion prediction information from a bitstream;
generating an affine motion vector predictor (MVP) candidate list including affine MVP candidates for a current block;
selecting one of the affine MVP candidates in the affine MVP candidate list based on an affine MVP candidate index in the motion prediction information;
deriving control point motion vector predictors (CPMVPs) for respective control points (CPs) of the current block based on the selected affine MVP candidate;
deriving control point motion vector differences (CPMVDs) for the respective CPs of the current block based on information on the CPMVDs for the respective CPs included in the obtained motion prediction information;
deriving control point motion vectors (CPMVs) for the respective CPs of the current block based on the CPMVPs and the CPMVDs;
deriving prediction samples for the current block based on the CPMVs; and
generating reconstructed samples for the current block based on the derived prediction samples,
wherein the affine MVP candidates in the affine MVP candidate list comprise a first affine MVP candidate and a second affine MVP candidate,
wherein the first affine MVP candidate is derived based on a first block being coded based on an affine motion model in a left block group including a bottom-left corner neighboring block and a left neighboring block of the current block,
wherein the second affine MVP candidate is derived based on a second block being coded based on the affine motion model in a top block group including a top-right corner neighboring block, a top neighboring block and a top-left corner neighboring block of the current block,
wherein the left block group excludes the top-right corner neighboring block, the top neighboring block and the top-left corner neighboring block, and
wherein the top block group excludes the bottom-left corner neighboring block and the left neighboring block.
2. The picture decoding method of claim 1 ,
wherein the motion prediction information includes information regarding a context index for the affine MVP candidate index,
wherein a value of the information regarding the context index is equal to 0 based on existence of at least one neighboring block for the current block coded according to the affine motion model,
wherein the value of the information on the context index is equal to 1 when no neighboring block for the current block coded according to the affine motion model exists, and
wherein the affine MVP candidate index is coded based on a context adaptive binary arithmetic coding by using a context model indicated by the context index.
3. A picture encoding method performed by an encoding apparatus, the method comprising:
generating an affine motion vector predictor (MVP) candidate list including affine MVP candidates for a current block;
selecting one of the affine MVP candidates in the affine MVP candidate list;
deriving an affine MVP candidate index related to the selected affine MVP candidate;
deriving control point motion vector predictors (CPMVPs) for respective control points (CPs) of the current block based on the selected affine MVP candidate;
deriving control point motion vectors (CPMVs) for the respective CPs of the current block;
deriving control point motion vector differences (CPMVDs) for the respective CPs of the current block based on the CPMVPs and the CPMVs for the respective CPs;
deriving prediction samples for the current block based on the CPMVs;
deriving residual samples for the current block based on the derived prediction samples; and
encoding information related to the affine MVP candidate index, information on the derived CPMVDs and residual information on the residual samples,
wherein the affine MVP candidates in the affine MVP candidate list comprise a first affine MVP candidate and a second affine MVP candidate,
wherein the first affine MVP candidate is derived based on a first block being coded based on an affine motion model in a left block group including a bottom-left corner neighboring block and a left neighboring block of the current block,
wherein the second affine MVP candidate is derived based on a second block being coded based on the affine motion model in a top block group including a top-right corner neighboring block, a top neighboring block and a top-left corner neighboring block of the current block,
wherein the left block group excludes the top-right corner neighboring block, the top neighboring block and the top-left corner neighboring block, and
wherein the top block group excludes the bottom-left corner neighboring block and the left neighboring block.
4. The picture encoding method of claim 3 , further comprising:
encoding information regarding a context index for the affine MVP candidate index,
wherein a value of the information regarding the context index is equal to 0 based on existence of at least one neighboring block for the current block coded according to the affine motion model,
wherein the value of the information on the context index is equal to 1 when no neighboring block for the current block coded according to the affine motion model exists, and
wherein the affine MVP candidate index is coded based on a context adaptive binary arithmetic coding by using a context model indicated by the context index.
5. A non-transitory computer-readable storage medium storing a bitstream generated by a picture encoding method, the method comprising:
generating an affine motion vector predictor (MVP) candidate list including affine MVP candidates for a current block;
selecting one of the affine MVP candidates in the affine MVP candidate list;
deriving an affine MVP candidate index related to the selected affine MVP candidate;
deriving control point motion vector predictors (CPMVPs) for respective control points (CPs) of the current block based on the selected affine MVP candidate;
deriving control point motion vectors (CPMVs) for the respective CPs of the current block;
deriving control point motion vector differences (CPMVDs) for the respective CPs of the current block based on the CPMVPs and the CPMVs for the respective CPs;
deriving prediction samples for the current block based on the CPMVs;
deriving residual samples for the current block based on the derived prediction samples; and
encoding information related to the affine MVP candidate index, information on the derived CPMVDs and residual information on the residual samples to generate the bitstream,
wherein the affine MVP candidates in the affine MVP candidate list comprise a first affine MVP candidate and a second affine MVP candidate,
wherein the first affine MVP candidate is derived based on a first block being coded based on an affine motion model in a left block group including a bottom-left corner neighboring block and a left neighboring block of the current block,
wherein the second affine MVP candidate is derived based on a second block being coded based on the affine motion model in a top block group including a top-right corner neighboring block, a top neighboring block and a top-left corner neighboring block of the current block,
wherein the left block group excludes the top-right corner neighboring block, the top neighboring block and the top-left corner neighboring block, and
wherein the top block group excludes the bottom-left corner neighboring block and the left neighboring block.
6. The non-transitory computer-readable storage medium of claim 5 , wherein the method further comprises:
encoding information regarding a context index for the affine MVP candidate index,
wherein a value of the information regarding the context index is equal to 0 based on existence of at least one neighboring block for the current block coded according to the affine motion model,
wherein the value of the information on the context index is equal to 1 when no neighboring block for the current block coded according to the affine motion model exists, and
wherein the affine MVP candidate index is coded based on a context adaptive binary arithmetic coding by using a context model indicated by the context index.