Method and device for decoding image in image coding system

An image encoding method performed by an encoding apparatus, the method comprising:

• deriving motion information of a first target block in a first picture;
• deriving modified motion information for the first target block based on the motion information and bi-prediction;
• when a merge mode is applied to a second target block in a second picture, configuring a merge candidate list for the second target block based on spatial neighboring blocks and a temporal neighboring block of the second target block;
• selecting one of merge candidates constituting the merge candidate list based on the merge index;
• generating merge index for the second target block indicating the selected merge candidate;
• and
• encoding image information including the merge index,
• wherein the merge candidates include spatial motion information candidates and a temporal motion information candidate, wherein the temporal motion information candidate is derived based on the temporal neighboring block,
• wherein when the temporal neighboring block is related to the first target block, the temporal motion information candidate is derived based on the modified motion information of the first target block,
• wherein when a size of a first target block is greater than a preconfigured size, the first target block is partitioned into a plurality of sub blocks and a sub reference block in a unit of each sub block in a reference picture is derived,
• wherein modified motion vectors for the sub blocks in the first target block are derived based on the derived sub reference blocks,
• wherein the modified motion information for the first target block includes at least one of the modified motion vectors for the sub blocks.

A non-transitory computer-readable storage medium storing encoded video data comprising information on instructions, when executed,

• deriving motion information of a first target block in a first picture;
• deriving modified motion information for the first target block based on the motion information and bi-prediction;
• when a merge mode is applied to a second target block in a second picture, configuring a merge candidate list for the second target block based on spatial neighboring blocks and a temporal neighboring block of the second target block;
• receiving a merge index for the second target block;
• selecting one of merge candidates constituting the merge candidate list based on the merge index;
• deriving motion information of the second target block based on the selected merge candidate;
• and
• generating prediction sample for the second target block by performing inter prediction based on the motion information of the second target block,
• wherein the merge candidates include spatial motion information candidates and a temporal motion information candidate, wherein the temporal motion information candidate is derived based on the temporal neighboring block,
• wherein when the temporal neighboring block is related to the first target block, the temporal motion information candidate is derived based on the modified motion information of the first target block,
• wherein when a size of a first target block is greater than a preconfigured size, the first target block is partitioned into a plurality of sub blocks and a sub reference block in a unit of each sub block in a reference picture is derived,
• wherein modified motion vectors for the sub blocks in the first target block are derived based on the derived sub reference blocks,
• wherein the modified motion information for the first target block includes at least one of the modified motion vectors for the sub blocks.