Method and device for video coding using local illumination compensation-based prediction in geometric partitioning mode

A method for encoding a current block by a video encoding apparatus, the method comprising:

• determining a geometric partitioning mode index;
• partitioning the current block into partitioned blocks according to the geometric partitioning mode index;
• generating reference blocks of the partitioned blocks by using motion information, the reference blocks including a first reference block and a second reference block;
• generating a prediction block of the current block by blending the reference blocks of the partitioned blocks;
• and
• checking, at a top side and a left side of the current block, availability of neighboring reference pixels of the current block,
• wherein the method further comprises, when the neighboring reference pixels of the current block are available,
• checking, at a top side and a left side of each reference block of the reference blocks, availability of neighboring reference pixels of each reference block, and
• wherein the method further comprises, when the neighboring reference pixels of each reference block of the reference blocks are available in whole or in part,
• generating, by blending neighboring pixels of the reference blocks, neighboring reference pixels of the blended prediction block;
• deriving parameters of a linear relation between the neighboring reference pixels of the current block and the neighboring reference pixels of the blended prediction block;
• and
• generating a first final prediction block of the current block by applying a local illumination compensation to the blended prediction block based on the parameters of the linear relation.

A method for providing video data to a video decoding apparatus, the method comprising:

• encoding the video data into a bitstream;
• and
• transmitting the bitstream to the video decoding apparatus,
• wherein encoding the video data comprises:
• determining a geometric partitioning mode index;
• partitioning a current block into partitioned blocks according to the geometric partitioning mode index;
• generating reference blocks of the partitioned blocks by using motion information, the reference blocks including a first reference block and a second reference block;
• generating a prediction block of the current block by blending the reference blocks of the partitioned blocks;
• and
• checking, at a top side and a left side of the current block, availability of neighboring reference pixels of the current block,
• wherein the encoding of the video data further comprises, when the neighboring reference pixels of the current block are available,
• checking, at a top side and a left side of each reference block of the reference blocks, availability of neighboring reference pixels of each reference block, and
• wherein the encoding of the video data further comprises, when the neighboring reference pixels of each reference block of the reference blocks are available in whole or in part,
• generating, by blending neighboring pixels of the reference blocks, neighboring reference pixels of the blended prediction block;
• deriving parameters of a linear relation between the neighboring reference pixels of the current block and the neighboring reference pixels of the blended prediction block;
• and
• generating a final prediction block of the current block by applying a local illumination compensation to the blended prediction block based on the parameters of the linear relation.