Video coding method and device using motion compensation of decoder side
A video coding method and device, according to the present invention, determine whether motion compensation is performed by sub-block unit, determine a search area for motion compensation of a current block, calculate a plurality of SAD candidates with respect to the search area, derive delta motion information of the current block on the basis of the plurality of SAD candidates, and can compensate for motion information of the current block by using pre-generated motion information and the delta motion information of the current block.
1. A method of decoding an image, comprising:
determining, based on a pre-defined condition, whether to perform motion modification on a current block in the image;
determining whether the motion modification is performed in units of sub-blocks when it is determined that the motion modification is performed on the current block;
determining a search region for the motion modification of the current block;
deriving delta motion information of the current block based on a plurality of SAD candidates in the search region; and
modifying initial motion information of the current block by using the delta motion information,
wherein whether the motion modification is performed in units of sub-blocks is determined based on a comparison result between a size of the current block and a pre-defined threshold size,
wherein, in response to a case where the motion modification is performed in units of sub-blocks, the current block is divided into a plurality of sub-blocks having the pre-defined threshold size,
wherein the motion modification of the current block is performed only when a picture order count (POC) difference between a current picture including the current block and an L0 reference picture is the same as a POC difference between the current picture and an L1 reference picture,
wherein the L0 reference picture and the L1 reference picture have different temporal directions,
wherein, in response to a case where an initial SAD is less than a specific value, a search for the plurality of SAD candidates in the search region is not performed even though it is determined that the motion modification is performed on the current block, and
wherein the initial SAD is derived based on reference blocks specified by the initial motion information of the current block.
2. The method of claim 1 , wherein the search region is determined as a region including at least one of a reference block or a neighboring region of the reference block, and
wherein the neighboring region is N sample lines neighboring to a boundary of the reference block.
3. The method of claim 1 , wherein each of the SAD candidates is determined as an SAD value between an L0 block and an L1 block, and
wherein the L0 block is determined based on a position of an L0 reference block of the current block and a predetermined offset and the L1 block is determined based on a position of an L1 reference block of the current block and the predetermined offset.
4. The method of claim 3 , wherein the SAD value is calculated using samples belonging to even rows, not samples belonging to all rows of the L0 block and the L1 block.
5. The method of claim 3 , wherein the predetermined offset includes at least one of a non-directional offset or a directional offset, and
wherein the directional offset includes an offset for at least one direction of a left, a right, a top, a bottom, a top-left, a top-right, a bottom-left or a bottom-right.
6. The method of claim 1 , wherein a first weight and a second weight for bi-prediction of the current block are applied to the L0 reference picture and the L1 reference picture, respectively, and
wherein the motion modification of the current block is performed only when the first weight is equal to the second weight.
7. The method of claim 6 , wherein the motion modification of the current block is performed only when a width and a height of the current block are greater than or equal to 8, respectively, and a number of samples included in the current block is greater than or equal to 128.
8. The method of claim 1 , wherein the pre-defined threshold size is equal to 16×16.
9. The method of claim 1 , wherein the motion modification of the current block is performed only when a combined prediction mode is not applied to the current block, and
wherein the combined prediction mode is a prediction mode combining intra prediction and inter prediction.
10. The method of claim 1 , wherein, in response to a case where it is determined that the motion modification is not performed in units of sub-blocks, the specific value is determined using a width and a height of the current block, and
wherein, in response to a case where it is determined that the motion modification is performed in units of sub-blocks, the specific value is determined using a width and a height of a sub-block belonging to the current block.
11. A method of encoding an image, the method comprising:
determining, based on a pre-defined condition, whether to perform motion modification on a current block in the image;
determining whether the motion modification is performed in units of sub-blocks when it is determined that the motion modification is performed on the current block;
determining a search region for the motion modification of the current block;
deriving delta motion information of the current block based on a plurality of SAD candidates in the search region; and
modifying initial motion information of the current block by using the delta motion information,
wherein whether the motion modification is performed in units of sub-blocks is determined based on a comparison result between a size of the current block and a pre-defined threshold size,
wherein, in response to a case where the motion modification is performed in units of sub-blocks, the current block is divided into a plurality of sub-blocks having the pre-defined threshold size,
wherein the motion modification of the current block is performed only when a picture order count (POC) difference between a current picture including the current block and an L0 reference picture is the same as a POC difference between the current picture and an L1 reference picture,
wherein the L0 reference picture and the L1 reference picture have different temporal directions,
wherein, in response to a case where an initial SAD is less than a specific value, a search for the plurality of SAD candidates in the search region is not performed even though it is determined that the motion modification is performed on the current block, and
wherein the initial SAD is derived based on reference blocks specified by the initial motion information of the current block.
12. A non-transitory computer-readable storage medium having instructions stored thereon that, when executed, cause one or more processor to:
determining, based on a pre-defined condition, whether to perform motion modification on a current block;
determining whether the motion modification is performed in units of sub-blocks when it is determined that the motion modification is performed on the current block;
determining a search region for the motion modification of the current block;
deriving delta motion information of the current block based on a plurality of SAD candidates in the search region; and
modifying initial motion information of the current block by using the delta motion information,
wherein whether the motion modification is performed in units of sub-blocks is determined based on a comparison result between a size of the current block and a pre-defined threshold size,
wherein, in response to a case where the motion modification is performed in units of sub-blocks, the current block is divided into a plurality of sub-blocks having the pre-defined threshold size,
wherein the motion modification of the current block is performed only when a picture order count (POC) difference between a current picture including the current block and an L0 reference picture is the same as a POC difference between the current picture and an L1 reference picture,
wherein the L0 reference picture and the L1 reference picture have different temporal directions,
wherein, in response to a case where an initial SAD is less than a specific value, a search for the plurality of SAD candidates in the search region is not performed even though it is determined that the motion modification is performed on the current block, and
wherein the initial SAD is derived based on reference blocks specified by the initial motion information of the current block.
13. A device having instructions for transmitting a bitstream generated by an encoding method,
wherein the encoding method comprises:
determining, based on a pre-defined condition, whether to perform motion modification on a current block;
determining whether the motion modification is performed in units of sub-blocks when it is determined that the motion modification is performed on the current block;
determining a search region for the motion modification of the current block;
deriving delta motion information of the current block based on a plurality of SAD candidates in the search region; and
modifying initial motion information of the current block by using the delta motion information,
wherein whether the motion modification is performed in units of sub-blocks is determined based on a comparison result between a size of the current block and a pre-defined threshold size,
wherein, in response to a case where the motion modification is performed in units of sub-blocks, the current block is divided into a plurality of sub-blocks having the pre-defined threshold size,
wherein the motion modification of the current block is performed only when a picture order count (POC) difference between a current picture including the current block and an L0 reference picture and a POC difference between the current picture and an L1 reference picture are the same,
wherein the L0 reference picture and the L1 reference picture have different temporal directions,
wherein, in response to a case where an initial SAD is less than a specific value, a search for the plurality of SAD candidates in the search region is not performed even though it is determined that the motion modification is performed on the current block, and
wherein the initial SAD is derived based on reference blocks specified by the initial motion information of the current block.