Inter-prediction method and apparatus for same
An inter-prediction method according to the present invention comprises the steps of: deriving motion information of a current block; and generating a prediction block for the current block on the basis of the derived motion information. According to the present invention, computational complexity can be reduced and encoding efficiency can be improved.
1. A method for decoding a video signal, the method comprising:
determining a prediction mode for a current block as an inter prediction mode;
generating, based on the determination, a merge candidate list of the current block by using neighboring blocks spatially adjacent to the current block and neighboring blocks temporally adjacent to the current block, each of the neighboring blocks spatially adjacent to the current block and the neighboring blocks temporally adjacent to the current block having motion information;
deriving, based on a merge index, motion information of a current block from the merge candidate list, the merge index specifying a merge candidate included in the merge candidate list;
generating a prediction block for the current block based on the derived motion information;
obtaining a residual block for the current block by decoding an input bitstream;
generating a reconstruction block for the current block based on the residual block and the prediction block; and
applying in-loop filter on the reconstruction block, wherein the in-loop filter is at least one of a deblocking filter and an SAO (Sample Adaptive Offset) filter,
wherein the derived motion information includes prediction direction information and both of L0 motion information and L1 motion information,
wherein, when the prediction direction information of the current block indicates a bi-prediction and a size of the current block is smaller than a pre-determined size, the prediction direction information indicating the bi-prediction is changed to indicate a uni-prediction and the prediction block of the current block is generated by performing motion compensation based on only the L0 motion information among the L0 motion information and the L1 motion information,
wherein, when the prediction direction information indicates the bi-prediction and the size of the current block is equal to or greater than the pre-determined size, the prediction direction information indicating the bi-prediction is not changed to indicate the uni-prediction and the prediction block of the current block is generated by performing the motion compensation based on both the L0 motion information and the L1 motion information, and
wherein the pre-determined size is representative of one of 8×8, 16×16, or 32×32.
2. The method of claim 1 , wherein the L0 motion information includes an L0 motion vector and an L0 reference picture number, the L0 reference picture number is a number assigned to an L0 reference picture according to a picture order count (POC) order, the L1 motion information includes an L1 motion vector and an L1 reference picture number, and the L1 reference picture number is a number assigned to an L1 reference picture according to the POC order.
3. The method of claim 2 , wherein generating the prediction block comprising:
generating an L0 motion compensated block by performing L0 motion compensation on the current block based on the L0 motion information;
generating an L1 motion compensated block by performing L1 motion compensation on the current block based on the L1 motion information; and
generating the prediction block by performing a weighting average on the L0 motion compensated block and the L1 motion compensated block.
4. The method of claim 1 , further comprising:
determining whether default weighting prediction is applied to the current block or non-default weighting prediction is applied thereto,
wherein the prediction block is generated by performing one of the default weighting prediction and the non-default weighting prediction according to the determination result.
5. The method of claim 4 , wherein whether the default weighting prediction is applied to the current block or the non-default weighting prediction is applied thereto is indicated by a weighting prediction index.
6. A method for encoding a video signal, the method comprising:
determining a prediction mode for a current block as an inter prediction mode;
generating, based on the determination, a merge candidate list of the current block by using neighboring blocks spatially adjacent to the current block and neighboring blocks temporally adjacent to the current block, each of the neighboring blocks spatially adjacent to the current block and the neighboring blocks temporally adjacent to the current block having motion information;
deriving, based on a merge index, motion information of a current block from the merge candidate list, the merge index specifying a merge candidate included in the merge candidate list;
generating a prediction block for the current block based on the derived motion information;
obtaining a residual block for the current block;
generating a reconstruction block for the current block based on the residual block and the prediction block; and
applying in-loop filter on the reconstruction block, wherein the in-loop filter is at least one of a deblocking filter and an SAO (Sample Adaptive Offset) filter,
wherein the derived motion information includes prediction direction information and both of L0 motion information and L1 motion information,
wherein, when the prediction direction information of the current block indicates a bi-prediction and a size of the current block is smaller than a pre-determined size, the prediction direction information indicating the bi-prediction is changed to indicate a uni-prediction and the prediction block of the current block is generated by performing motion compensation based on only the L0 motion information among the L0 motion information and the L1 motion information,
wherein, when the prediction direction information indicates the bi-prediction and the size of the current block is equal to or greater than the pre-determined size, the prediction direction information indicating the bi-prediction is not changed to indicate the uni-prediction and the prediction block of the current block is generated by performing the motion compensation based on both the L0 motion information and the L1 motion information, and
wherein the pre-determined size is representative of one of 8×8, 16×16, or 32×32.
7. The method of claim 6 , wherein the L0 motion information includes an L0 motion vector and an L0 reference picture number, the L0 reference picture number is a number assigned to an L0 reference picture according to a picture order count (POC) order, the L1 motion information includes an L1 motion vector and an L1 reference picture number, and the L1 reference picture number is a number assigned to an L1 reference picture according to the POC order.
8. The method of claim 7 , wherein generating the prediction block comprising:
generating an L0 motion compensated block by performing L0 motion compensation on the current block based on the L0 motion information;
generating an L1 motion compensated block by performing L1 motion compensation on the current block based on the L1 motion information; and
generating the prediction block by performing a weighting average on the L0 motion compensated block and the L1 motion compensated block.
9. The method of claim 6 , further comprising:
determining whether default weighting prediction is applied to the current block or non-default weighting prediction is applied thereto,
wherein the prediction block is generated by performing one of the default weighting prediction and the non-default weighting prediction according to the determination result.
10. The method of claim 9 , wherein whether the default weighting prediction is applied to the current block or the non-default weighting prediction is applied thereto is indicated by a weighting prediction index.
11. A non-transitory computer-readable recoding-medium storing a bitstream generated by an encoding method, the encoding method comprising:
determining a prediction mode for a current block as an inter prediction mode;
generating, based on the determination, a merge candidate list of the current block by using neighboring blocks spatially adjacent to the current block and neighboring blocks temporally adjacent to the current block, each of the neighboring blocks spatially adjacent to the current block and the neighboring blocks temporally adjacent to the current block having motion information;
deriving, based on a merge index, motion information of a current block from the merge candidate list, the merge index specifying a merge candidate included in the merge candidate list;
generating a prediction block for the current block based on the derived motion information;
obtaining a residual block for the current block;
generating a reconstruction block for the current block based on the residual block and the prediction block; and
applying in-loop filter on the reconstruction block, wherein the in-loop filter is at least one of a deblocking filter and an SAO (Sample Adaptive Offset) filter,
wherein the derived motion information includes prediction direction information and both of L0 motion information and L1 motion information,
wherein, when the prediction direction information of the current block indicates a bi-prediction and a size of the current block is smaller than a pre-determined size, the prediction direction information indicating the bi-prediction is changed to indicate a uni-prediction and the prediction block of the current block is generated by performing motion compensation based on only the L0 motion information among the L0 motion information and the L1 motion information,
wherein, when the prediction direction information indicates the bi-prediction and the size of the current block is equal to or greater than the pre-determined size, the prediction direction information indicating the bi-prediction is not changed to indicate the uni-prediction and the prediction block of the current block is generated by performing the motion compensation based on both the L0 motion information and the L1 motion information, and
wherein the pre-determined size is representative of one of 8×8, 16×16, or 32×32.