Method for encoding/decoding image signal and device therefor
An image decoding method according to the present invention may comprise the steps of: generating an affine mergence candidate list for a current block; specifying one of multiple affine mergence candidates contained in the affine mergence candidate list; deriving affine seed vectors of the specified mergence candidate; deriving a sub-block motion vector for a sub-block in the current block by using the affine seed vectors, wherein the sub-block is smaller in size than the current block; deriving offset data for the sub-block; and generating a sub-prediction block for the sub-block on the basis of the affine vector and the offset data.
1. A method of decoding a video, the method comprising:
determining control point motion vectors for a current block;
generating an affine merge candidate list for the current block;
specifying one of a plurality of affine merge candidates included in the affine merge candidate list;
deriving a motion vector for a sub-block in the current block based on the control point motion vectors and the specified affine merge candidate, the sub-block being a smaller region than the current block;
deriving offset data for the sub-block; and
generating prediction samples of the sub-block based on the motion vector and the offset data,
wherein a prediction sample in the sub-block is obtained based on a first prediction sample derived based on the motion vector of the sub-block and an offset value, the offset value being derived based a differential vector for the prediction sample.
2. The method of claim 1 , wherein the offset data is an array of differential vectors each of which corresponds to each of prediction samples in the sub-block.
3. The method of claim 1 , wherein the offset value is derived based on a horizontal directional gradient and a vertical directional gradient for the prediction sample.
4. The method of claim 1 , wherein whether the offset data is used in generating the prediction samples is determined based on whether the control point motion vectors are mutually the same or not.
5. A method of encoding a video, the method comprising:
determining control point motion vectors for a current block;
generating an affine merge candidate list for the current block;
specifying one of a plurality of affine merge candidates included in the affine merge candidate list;
deriving a motion vector for a sub-block in the current block based on the control point motion vectors and the specified affine merge candidate, the sub-block being a smaller region than the current block;
deriving offset data for the sub-block; and
generating prediction samples of the sub-block based on the motion vector and the offset data,
wherein a prediction sample in the sub-block is obtained based on a first prediction sample derived based on the motion vector of the sub-block and an offset value, the offset value being derived based a differential vector for the prediction sample.
6. The method of claim 5 , wherein the offset data is an array for differential vectors each of which corresponds to each of prediction samples in the sub-block.
7. The method of claim 5 , wherein the offset value is derived based on a horizontal directional gradient and a vertical directional gradient for the prediction sample.
8. The method of claim 5 , wherein whether the offset data is used in generating the prediction samples is determined based on whether the control point motion vectors are mutually the same or not.
9. A non-transitory computer readable medium having stored thereon a compressed video data, the compressed video data comprising:
index information specifying one of a plurality of affine merge candidates included in an affine merge candidate list,
wherein a motion vector for a sub-block in the current block is derived based on control point motion vectors of the current block and the specified affine merge candidate, the sub-block being a smaller region than the current block;
wherein prediction samples of the sub-block are generated based on the motion vector and offset data for the sub-block, and
wherein a prediction sample in the sub-block is obtained based on a first prediction sample derived based on the motion vector of the sub-block and an offset value, the offset value being derived based a differential vector for the prediction sample.
10. The non-transitory computer readable medium of claim 9 , wherein whether the offset data is used in generating the prediction samples is determined based on whether the control point motion vectors are mutually the same or not.