Picture prediction method and related apparatus

An encoding apparatus, comprising:

• a non-transitory memory having processor-executable instructions stored thereon;
• and
• a processor, coupled to the non-transitory memory, configured to execute the processor-executable instructions to facilitate:
• determining motion vector predictors (MVPs) of K pixel samples in a current picture block, wherein K is an integer greater than 1;
• obtaining respective motion vectors of the K pixel samples via motion estimation;
• obtaining motion vector differences (MVDs) of the K pixel samples based on the respective motion vectors of the K pixel samples and corresponding MVPs;
• based on the MVDs of the K pixel samples all being zero, writing a flag into a bitstream, wherein the value of the flag indicates that the MVDs of the K pixel samples all being zero;
• calculating, based on a non-translational motion model , a motion vector of each pixel or each sub-block in the current picture block;
• and
• performing, based on the calculated motion vector of each pixel or each sub-block in the current picture block, pixel value prediction on the current picture block.
• wherein the non-translational motion model is represented by vx=a×x+b×y+e and vy=c×x+d×y+f, where (x,y) corresponds to the coordinates of a respective pixel or sub-block in the current picture block, and a, b, e, c, d and f are parameters of the non-translational motion model and are derived based on the respective motion vectors of the K pixel samples of the current picture block and a width and a height of the current picture bloc and wherein (vx, vy) represents a respective motion vector of a respective pixel or sub-block calculated based on the non-translational motion model.

A picture prediction method for a video decoding process, comprising:

• determining, by a decoding apparatus, motion vector predictors (MVPs) of K pixel samples in a current picture block, wherein K is an integer greater than 1;
• decoding a flag of the current picture block from a bitstream;
• based on the value of the flag indicating that the MVDs of the K pixel samples all being zero, determining the MVDs of the K pixel samples are all 0;
• and obtaining respective motion vectors of the K pixel samples based on the MVDs of the K pixel samples and corresponding MVPs of the K pixel samples;
• calculating, based on a non-translational motion model, a motion vector of each pixel or each sub-block in the current picture block;
• and
• performing, based on the calculated motion vector of each pixel or each sub-block in the current picture block, pixel value prediction on the current picture block.
• wherein the non-translational motion model is represented by vx=a×x+b×y+e and vy=c×x+d×y+f, where (x,y) corresponds to the coordinates of a respective pixel or sub-block in the current picture block, and a, b, e, c, d and f are parameters of the non-translational motion model and are derived based on the respective motion vectors of the K pixel samples of the current picture block and a width and a height of the current picture bloc and wherein (vx, vy) represents a respective motion vector of a respective pixel or sub-block calculated based on the non-translational motion model.

A decoding apparatus, comprising:

• a non-transitory memory having processor-executable instructions stored thereon;
• and
• a processor, coupled to the non-transitory memory, configured to execute the processor-executable instructions to facilitate:
• determining motion vector predictors (MVPs) of K pixel samples in a current picture block, wherein K is an integer greater than 1;
• decoding a flag of the current picture block from a bitstream;
• based on the value of the flag indicating that the MVDs of the K pixel samples all being zero, determining the MVDs of the K pixel samples are all 0;
• and obtaining respective motion vectors of the K pixel samples based on the MVDs of the K pixel samples and corresponding MVPs of the K pixel samples;
• calculating, based on a non-translational motion model, a motion vector of each pixel or each sub-block in the current picture block;
• and
• performing, based on the calculated motion vector of each pixel or each sub-block in the current picture block, pixel value prediction on the current picture block.
• wherein the non-translational motion model is represented by vx=a×x+b×y+e and vy=c×x+d×y+f, where (x,y) corresponds to the coordinates of a respective pixel or sub-block in the current picture block, and a, b, e, c, d and f are parameters of the non-translational motion model and are derived based on the respective motion vectors of the K pixel samples of the current picture block and a width and a height of the current picture bloc and wherein (vx, vy) represents a respective motion vector of a respective pixel or sub-block calculated based on the non-translational motion model.