Image encoding/decoding method and recording medium therefor
The present invention relates to a video encoding and decoding method. A video decoding method for the same may include: obtaining granularity information; determining a sharing region based on the granularity information; and generating a prediction block of a current block by using motion information of the sharing region when the current block corresponds to the sharing region.
1. A video decoding method, comprising:
determining a prediction mode of a current block, the prediction mode being determined as one of intra prediction and inter prediction;
in response to the prediction mode of the current block being the inter prediction, determining which of a merge mode and a motion vector prediction mode is applied to the current block;
in response to the motion vector prediction mode being applied to the current block, decoding from a bitstream, an inter prediction indicator for the current block, the inter prediction indicator indicating whether a uni-directional or bi-directional inter prediction is applied to the current block;
in response to the bi-directional inter prediction being applied to the current block, obtaining an L0 motion vector and an L1 motion vector of the current block;
wherein the L0 motion vector is derived by adding an L0 motion vector difference to an L0 motion vector candidate,
wherein the L0 motion vector candidate is specified by an L0 motion vector candidate index specifying the L0 motion vector candidate from an L0 motion vector candidate list,
wherein at least one of an L0 spatial motion vector candidate derived from a neighboring block adjacent to the current block or an L0 temporal motion vector candidate derived from a collocated block of the current block is included in the L0 motion vector candidate list,
wherein the L1 motion vector is derived by adding an L1 motion vector difference to an L1 motion vector candidate,
wherein the L1 motion vector candidate is specified by an L1 motion vector candidate index specifying the L1 motion vector candidate from an L1 motion vector candidate list, and
wherein at least one of an L1 spatial motion vector candidate derived from a neighboring block adjacent to the current block or an L1 temporal motion vector candidate derived from a collocated block of the current block is included in the L1 motion vector candidate list,
obtaining an L0 inter prediction block of the current block from an L0 reference picture based on the L0 motion vector;
obtaining an L1 inter prediction block of the current block from an L1 reference picture based on the L1 motion vector; and
generating a final inter prediction block of the current block based on a weighted sum of the L0 inter prediction block and the L1 inter prediction block,
wherein the weighted sum is performed by applying a first weighting factor to the L0 inter prediction block and applying a second weighting factor to the L1 inter prediction block,
wherein the first weighting factor and the second weighting factor are determined based on index information for the current block,
wherein the index information is explicitly signaled via the bitstream only when the inter prediction indicator of the current block indicates that the bi-directional inter prediction is applied to the current block,
wherein motion vector difference information is explicitly signaled only for the L0 motion vector difference but is not explicitly signaled for the L1 motion vector difference,
wherein a plurality of pre-defined weighting factor set candidates comprise a candidate in which at least one of the first weighting factor and the second weighting factor is a negative value, and
wherein a pair of the first weighting factor and the second weighting factor is determined as one, specified by the index information, of the pre-defined weighting factor set candidates.
2. The method of claim 1 , wherein the weight information of the current block is binarized with a truncated rice binarization method.
3. A video encoding method, comprising:
obtaining an intra predicted block of a current block by performing intra prediction for the current block;
obtaining an inter predicted block of the current block by performing inter prediction for the current block; and
determining a prediction mode of the current block by comparing a cost of the intra predicted block and a cost of the inter predicted block, the prediction mode being determined as one of the intra prediction and the inter prediction,
wherein in response to the prediction mode of the current block being the inter prediction, the method further comprises:
determining which of a merge mode and a motion vector prediction mode is applied to the current block; and
in response to the motion vector prediction mode being applied to the current block, encoding, into a bitstream, an inter prediction indicator for the current block, the inter prediction indicator indicating whether a uni-directional or bi-directional inter prediction is applied to the current block,
wherein in response to the bi-directional inter prediction being applied to the current block, the inter predicted block is obtained based on a weighted sum of an L0 inter prediction block and an L1 inter prediction block, the L0 inter prediction block being obtained from an L0 reference picture based on an L0 motion vector of the current block, the L1 inter prediction block being obtained from an L1 reference picture based on an L1 motion vector of the current block,
wherein an L0 motion vector difference is derived by subtracting an L0 motion vector candidate from the L0 motion vector,
wherein an L0 motion vector candidate index specifying the L0 motion vector candidate from an L0 motion vector candidate list is encoded,
wherein at least one of an L0 spatial motion vector candidate derived from a neighboring block adjacent to the current block or an L0 temporal motion vector candidate derived from a collocated block of the current block is included in the L0 motion vector candidate list,
wherein an L1 motion vector difference is derived by subtracting an L1 motion vector candidate from the L1 motion vector,
wherein an L1 motion vector candidate index specifying the L1 motion vector candidate from an L1 motion vector candidate list is encoded,
wherein at least one of an L1 spatial motion vector candidate derived from a neighboring block adjacent to the current block or an L1 temporal motion vector candidate derived from a collocated block of the current block is included in the L1 motion vector candidate list,
wherein the weighted sum is performed by applying a first weighting factor to the L0 inter prediction block and applying a second weighting factor to the L1 inter prediction block,
wherein index information, which is used to determine the first weighting factor and the second weighting factor at a decoder side, are encoded into the bitstream,
wherein the index information is encoded into the bitstream only when the inter prediction indicator of the current block is encoded with a value indicating that the bi-directional inter prediction is applied to the current block,
wherein motion vector difference information is explicitly encoded only for the L0 motion vector difference but is not explicitly encoded for the L1 motion vector difference,
wherein the index information specifies a weighting factor set candidate which corresponds to a pair of the first weighting factor and the second weighting factor, and
wherein a plurality of pre-defined weighting factor set candidates comprise a candidate in which at least one of the first weighting factor and the second weighting factor is a negative value.
4. The method of claim 3 , wherein the weight information of the current block is entropy coded using a truncated Rice binarization method.
5. A non-transitory recording medium storing a bit stream generated by a video encoding method, wherein the method includes:
obtaining an intra predicted block of a current block by performing intra prediction for the current block;
obtaining an inter predicted block of the current block by performing inter prediction for the current block; and
determining a prediction mode of the current block by comparing a cost of the intra predicted block and a cost of the inter predicted block, the prediction mode being determined as one of the intra prediction and the inter prediction,
wherein in response to the prediction mode of the current block being the inter prediction, the method further comprises:
determining which of a merge mode and a motion vector prediction mode is applied to the current block; and
in response to the motion vector prediction mode being applied to the current block, encoding, into a bitstream, an inter prediction indicator for the current block, the inter prediction indicator indicating whether a uni-directional or bi-directional inter prediction is applied to the current block,
wherein in response to the bi-directional inter prediction being applied to the current block, the inter predicted block is obtained based on a weighted sum of an L0 inter prediction block and an L1 inter prediction block, the inter L0 prediction block being obtained from an L0 reference picture based on an L0 motion vector of the current block, the L1 inter prediction block being obtained from an L1 reference picture based on an L1 motion vector of the current block,
wherein an L0 motion vector difference is derived by subtracting an L0 motion vector candidate from the L0 motion vector,
wherein an L0 motion vector candidate index specifying the L0 motion vector candidate from an L0 motion vector candidate list is encoded,
wherein at least one of an L0 spatial motion vector candidate derived from a neighboring block adjacent to the current block or an L0 temporal motion vector candidate derived from a collocated block of the current block is included in the L0 motion vector candidate list,
wherein an L1 motion vector difference is derived by subtracting an L1 motion vector candidate from the L1 motion vector,
wherein an L1 motion vector candidate index specifying the L1 motion vector candidate from an L1 motion vector candidate list is encoded,
wherein at least one of an L1 spatial motion vector candidate derived from a neighboring block adjacent to the current block or an L1 temporal motion vector candidate derived from a collocated block of the current block is included in the L1 motion vector candidate list,
wherein the weighted sum is performed by applying a first weighting factor to the L0 inter prediction block and applying a second weighting factor to the L1 inter prediction block,
wherein index information, which is used to determine the first weighting factor and the second weighting factor at a decoder side, are encoded into the bitstream,
wherein the index information is encoded into the bitstream only when the inter prediction indicator of the current block is encoded with a value indicating that the bi-directional inter prediction is applied to the current block,
wherein at least one of the first weighting factor and the second weighting factor is derived to be a negative value,
wherein motion vector difference information is explicitly encoded only for the L0 motion vector difference but is not explicitly encoded for the L1 motion vector difference,
wherein the index information specifies a weighting factor set candidate which corresponds to a pair of the first weighting factor and the second weighting factor, and
wherein a plurality of pre-defined weighting factor set candidates comprise a candidate in which at least one of the first weighting factor and the second weighting factor is a negative value.