Method and device for encoding or decoding image on basis of inter mode
In a method and a device for encoding or decoding an image according to the present invention, motion information for bidirectional prediction of a current block may be derived on the basis of an inter mode previously defined in the device for encoding or decoding an image, and inter prediction may be performed on the current block on the basis of the motion information, wherein the motion information for bidirectional prediction is adjusted to be motion information for unidirectional prediction according to the predefined inter mode.
1. A method of decoding an image signal with a decoding apparatus, comprising:
receiving, with the decoding apparatus, a bitstream including the image signal;
constructing, with the decoding apparatus, a merge candidate list of a current block, the merge candidate list including a plurality of merge candidates, the plurality of merge candidates including at least one of a spatial merge candidate, a temporal merge candidate, or a combined merge candidate;
deriving, with the decoding apparatus, motion information of the current block based on a merge candidate specified by a merge index among the plurality of merge candidates in the merge candidate list, the motion information including at least one of a prediction direction, a motion vector, or a reference picture index; and
performing, with the decoding apparatus, inter prediction on the current block based on the derived motion information,
wherein, in response to a case in which the current block uses geometric partitioning-based motion prediction, the prediction direction of the current block is restricted so that a first partition and a second partition resulting from geometric partitioning of the current block have only one of a L0 prediction direction or a L1 prediction direction, respectively,
wherein whether the first or second partition has the L0 prediction direction or not is variably determined based on the merge index specifying one of the plurality of merge candidates in the merge candidate list,
wherein reference picture indexes of L0 and L1 directions for the combined merge candidate are derived based on a reference picture index of one of a first merge candidate or a second merge candidate, and
wherein the one of the first merge candidate or the second merge candidate represents a merge candidate having a smallest candidate index among a candidate index of the first merge candidate and a candidate index of the second merge candidate.
2. The method of claim 1 , wherein a shape of at least one of the first and second partitions is triangular.
3. The method of claim 1 ,
wherein a motion vector of a L0 direction for the combined merge candidate is derived as an average of a motion vector of a L0 direction for the first merge candidate and a motion vector of a L0 direction for the second merge candidate, and
wherein a motion vector of a L1 direction for the combined merge candidate is derived as an average of a motion vector of a L1 direction for the first merge candidate and a motion vector of a L1 direction for the second merge candidate.
4. A method of encoding an image with an encoding apparatus, comprising:
obtaining, with the encoding apparatus, a prediction block of a current block based on motion information of the current block, the motion information being obtained from one of a plurality of merge candidates in a merge candidate list, the motion information including at least one of a prediction direction, a motion vector, or a reference picture index, the plurality of merge candidates including at least one of a spatial merge candidate, a temporal merge candidate, or a combined merge candidate;
obtaining, with the encoding apparatus, a residual block including residual data of the current block based on the prediction block and an original block of the current block; and
generating, with the encoding apparatus, a bitstream by encoding the residual data of the residual block,
wherein a merge index is encoded to specify the one of the plurality of merge candidates in the merge candidate list,
wherein, in response to a case in which the current block uses geometric partitioning-based motion prediction, the prediction direction of the current block is restricted so that a first partition and a second partition resulting from geometric partitioning of the current block have only one of a L0 prediction direction or a L1 prediction direction, respectively,
wherein whether the first or second partition has the L0 prediction direction or not is variably determined based on the merge index specifying one of the plurality of merge candidates in the merge candidate list,
wherein reference picture indexes of L0 and L1 directions for the combined merge candidate are derived based on a reference picture index of one of a first merge candidate or a second merge candidate, and
wherein the one of the first merge candidate or the second merge candidate represents a merge candidate having a smallest candidate index among a candidate index of the first merge candidate and a candidate index of the second merge candidate.
5. A non-transitory computer-readable medium containing instructions that, when executed by a processor, cause a device to process an image signal,
wherein a prediction block of a current block is obtained based on motion information of the current block,
wherein the motion information is obtained from one of a plurality of merge candidates in a merge candidate list,
wherein the motion information includes at least one of a prediction direction, a motion vector, or a reference picture index,
wherein the plurality of merge candidates includes at least one of a spatial merge candidate, a temporal merge candidate, or a combined merge candidate,
wherein a residual block including residual data of the current block is obtained based on the prediction block and an original block of the current block,
wherein encoding the residual data of the residual block is encoded,
wherein a merge index is encoded to specify the one of the plurality of merge candidates in the merge candidate list,
wherein, in response to a case in which the current block uses geometric partitioning-based motion prediction, the prediction direction of the current block is restricted so that a first partition and a second partition resulting from geometric partitioning of the current block have only one of a L0 prediction direction or a L1 prediction direction, respectively,
wherein whether the first or second partition has the L0 prediction direction or not is variably determined based on the merge index specifying one of the plurality of merge candidates in the merge candidate list,
wherein reference picture indexes of L0 and L1 directions for the combined merge candidate are derived based on a reference picture index of one of a first merge candidate or a second merge candidate, and
wherein the one of the first merge candidate or the second merge candidate represents a merge candidate having a smallest candidate index among a candidate index of the first merge candidate and a candidate index of the second merge candidate.