Image decoding method and apparatus according to block division structure in image coding system
Provided is a video decoding method performed by a decoding apparatus, which includes: obtaining split information for a target block from a bitstream; splitting the target block into a first sub-block and a second sub-block based on a split boundary indicated by the split information; deriving a first motion information candidate list for the first sub-block and a second motion information candidate list for the second sub-block based on the split information for the target block; performing inter prediction of the first sub-block based on the first motion information candidate list; and performing inter prediction of the second sub-block based on the second motion information candidate list, in which the first sub-block and the second sub-block are non-rectangular blocks, and the first motion information candidate list for the first sub-block is different from the second motion information candidate list for the second sub-block.
1. A video decoding method performed by a decoding apparatus, comprising:
obtaining residual information and split information for a target block from a bitstream;
splitting the target block into a first sub-partition and a second sub-partition based on a split boundary indicated by the split information;
deriving a first motion information candidate list for the first sub-partition and a second motion information candidate list for the second sub-partition;
deriving prediction samples in the first sub-partition based on the first motion information candidate list;
deriving prediction samples in the second sub-partition based on the second motion information candidate list;
deriving residual samples of the target block based on the residual information; and
generating a reconstructed picture based on the prediction samples in the first sub-partition, the prediction samples in the second sub-partition and the residual samples,
wherein the first sub-partition and the second sub-partition are non-rectangular partitions,
wherein the split information includes information on an angle of the split boundary and a distance between the split boundary and a center of the target block,
wherein the target block is split into the first sub-partition and the second sub-partition through the split boundary derived based on the information on the angle of the split boundary and the distance between the split boundary and the center of the target block, and
wherein based on a right height of the second sub-partition being RH, a size of the target block being N×N, and an x component of a top-left sample position of the target block being 0 and a y component of the top-left sample position being 0,
the second motion information candidate list includes at least one of a first spatial candidate indicating motion information of a first spatial neighboring block and a second spatial candidate indicating motion information of a second spatial neighboring block,
a location of the first spatial neighboring block is (N, N−1−RH), and
a location of the second spatial neighboring block is (N−1, N−1−RH).
2. The video decoding method of claim 1 , wherein based on the split boundary crossing an upper boundary and a left boundary of the target block, an upper width of the first sub-partition being UW, a left height of the first sub-partition being LH, and an x component of a top-left sample position of the target block being 0 and a y component of the top-left sample position being 0,
the first motion information candidate list includes at least one of a first spatial candidate indicating motion information of a first spatial neighboring block, a second spatial candidate indicating motion information of a second spatial neighboring block, a third spatial candidate indicating motion information of a third spatial neighboring block, a fourth spatial candidate indicating motion information of a fourth spatial neighboring block, and a temporal candidate indicating motion information of a temporal neighboring block in a co-located picture,
a location of the first spatial neighboring block is (−1, LH),
a location of the second spatial neighboring block is (−1, LH−1),
a location of the third spatial neighboring block is (UW, −1),
a location of the fourth spatial neighboring block is (UW−1, −1), and
a location of the temporal neighboring block may be (0, LH−1).
3. The video decoding method of claim 1 , wherein based on the split boundary crossing a right boundary and a lower boundary of the target block, a lower width of the first sub-partition being DW, a left height of the first sub-partition being LH, and an x component of a top-left sample position of the target block being 0 and a y component of the top-left sample position being 0, and
a location of a temporal neighboring block is (DW−1, LH−1).
4. The video decoding method of claim 1 , wherein based on the split boundary crossing a right boundary and a lower boundary of the target block, a lower width of the second sub-partition being DW, a right height of the second sub-partition being RH, a size of the target block being N x N, and an x component of a top-left sample position of the target block being 0 and a y component of the top-left sample position being 0,
the second motion information candidate list includes at least one of a third spatial candidate indicating motion information of a third spatial neighboring block and a fourth spatial candidate indicating motion information of a fourth spatial neighboring block,
a location of the third spatial neighboring block is (N−1−DW, N), and
a location of the fourth spatial neighboring block is (N−1−DW, N−1).
5. The video decoding method of claim 1 , wherein based on the split boundary crossing an upper boundary and a right boundary of the target block, an upper width of the first sub-partition being UW, and an x component of a top-left sample position of the target block being 0 and a y component of the top-left sample position being 0,
the first motion information candidate list includes at least one of a first spatial candidate indicating motion information of a first spatial neighboring block and a second spatial candidate indicating motion information of a second spatial neighboring block,
a location of the first spatial neighboring block is (UW, −1), and
a location of the second spatial neighboring block is (UW−1, −1).
6. The video decoding method of claim 1 , wherein based on the split boundary crossing an upper boundary and a right boundary of the target block, an upper width of the second sub-partition being UW, a size of the target block being N×N, and an x component of a top-left sample position of the target block being 0 and a y component of the top-left sample position being 0,
the second motion information candidate list includes a temporal candidate indicating motion information of a temporal neighboring block in a co-located picture,
a location of the temporal neighboring block is (N, N).
7. The video decoding method of claim 1 , wherein based on the split boundary crossing a left boundary and a lower boundary of the target block, a left height of the first sub-partition being LH, a lower width of the first sub-partition being DW, a size of the target block being N×N, and an x component of a top-left sample position of the target block being 0 and a y component of the top-left sample position being 0,
the first motion information candidate list includes a temporal candidate indicating motion information of a temporal neighboring block in a co-located picture, and
a location of the temporal neighboring block is (DW−1, N−1).
8. The video decoding method of claim 1 , wherein based on the split boundary crossing an upper boundary and a lower boundary of the target block, an upper width of the first sub-partition being UW, a lower width of the first sub-partition being DW, a size of the target block being N x N, and an x component of a top-left sample position of the target block being 0 and a y component of the top-left sample position being 0,
the first motion information candidate list includes at least one of a first spatial candidate indicating motion information of a first spatial neighboring block, a second spatial candidate indicating motion information of a second spatial neighboring block, and a temporal candidate indicating motion information of a temporal neighboring block in a co-located picture,
a location of the first spatial neighboring block is (UW, −1),
a location of the second spatial neighboring block is (UW−1, −1), and
a location of the temporal neighboring block is (DW−1, N−1).
9. The video decoding method of claim 1 , wherein based on the split boundary crossing an upper boundary and a lower boundary of the target block, an upper width of the second sub-partition being UW, a lower width of the second sub-partition being DW, a size of the target block being N×N, and an x component of a top-left sample position of the target block being 0 and a y component of the top-left sample position being 0,
the second motion information candidate list includes at least one of a third spatial candidate indicating motion information of a third spatial neighboring block, and a temporal candidate indicating motion information of a temporal neighboring block in a co-located picture,
a location of the third spatial neighboring block is (N−1−UW, −1), and
a location of the temporal neighboring block is (N, N).
10. The video decoding method of claim 1 , wherein based on the split boundary crossing a left boundary and a right boundary of the target block, a left height of the first sub-partition being LH, and an x component of a top-left sample position of the target block being 0 and a y component of the top-left sample position being 0,
the first motion information candidate list includes at least one of a first spatial candidate indicating motion information of a first spatial neighboring block, a second spatial candidate indicating motion information of a second spatial neighboring block, and a temporal candidate indicating motion information of a temporal neighboring block in a co-located picture,
a location of the first spatial neighboring block is (−1, LH),
a location of the second spatial neighboring block is (−1, LH−1), and
a location of the temporal neighboring block may be (0, LH−1).
11. A video encoding method performed by an encoding apparatus, comprising:
splitting a target block into a first sub-partition and a second sub-partition;
deriving a first motion information candidate list for the first sub-partition and a second motion information candidate list for the second sub-partition;
deriving prediction samples in the first sub-partition based on the first motion information candidate list;
deriving prediction samples in the second sub-partition based on the second motion information candidate list;
deriving residual samples of the target block based on the prediction samples in the first sub-partition and the prediction samples in the second sub-partition; and
encoding a bitstream including residual information and split information for the target block,
wherein the first sub-partition and the second sub-partition are non-rectangular partitions, and
wherein the split information includes information on an angle of a split boundary and a distance between the split boundary and a center of the target block,
wherein the target block is split into the first sub-partition and the second sub-partition through the split boundary derived based on the information on the angle of the split boundary and the distance between the split boundary and the center of the target block, and
wherein based on a right height of the second sub-partition being RH, a size of the target block being N×N, and an x component of a top-left sample position of the target block being 0 and a y component of the top-left sample position being 0,
the second motion information candidate list includes at least one of a first spatial candidate indicating motion information of a first spatial neighboring block and a second spatial candidate indicating motion information of a second spatial neighboring block,
a location of the first spatial neighboring block is (N, N−1−RH), and
a location of the second spatial neighboring block is (N−1, N−1−RH).
12. A non-transitory computer-readable storage medium storing a bitstream generated by a method, the method comprising:
splitting a target block into a first sub-partition and a second sub-partition;
deriving a first motion information candidate list for the first sub-partition and a second motion information candidate list for the second sub-partition;
deriving prediction samples in the first sub-partition based on the first motion information candidate list;
deriving prediction samples in the second sub-partition based on the second motion information candidate list;
deriving residual samples of the target block based on the prediction samples in the first sub-partition and the prediction samples in the second sub-partition; and
generating the bitstream including residual information and split information for the target block,
wherein the first sub-partition and the second sub-partition are non-rectangular partitions, and
wherein the split information includes information on an angle of a split boundary and a distance between the split boundary and a center of the target block,
wherein the target block is split into the first sub-partition and the second sub-partition through the split boundary derived based on the information on the angle of the split boundary and the distance between the split boundary and the center of the target block, and
wherein based on a right height of the second sub-partition being RH, a size of the target block being N×N, and an x component of a top-left sample position of the target block being 0 and a y component of the top-left sample position being 0,
the second motion information candidate list includes at least one of a first spatial candidate indicating motion information of a first spatial neighboring block and a second spatial candidate indicating motion information of a second spatial neighboring block,
a location of the first spatial neighboring block is (N, N−1−RH), and
a location of the second spatial neighboring block is (N−1, N−1−RH).