Decoder-side motion vector refinement (dmvr) process method and apparatus

An encoding apparatus, comprising:

• a memory having processor-executable instructions stored thereon;
• and
• one or more processors, coupled to the memory, configured to execute the processor-executable instructions to cause the encoder to perform steps of:
• obtaining initial motion vectors of a current block of a video frame, wherein the initial motion vectors comprise a first motion vector and a second motion vector;
• obtaining interpolated samples corresponding to two reference blocks of the current coding block based on the initial motion vectors from a bilinear interpolation, wherein the two reference blocks correspond to the first motion vector and the second motion vector, respectively;
• computing a sum of absolute differences (SAD) between the two reference blocks using a subset of the interpolated samples;
• determining the SAD is greater than or equal to a threshold value, wherein the threshold value is determined based on a size of the current block;
• performing remaining DMVR process steps;
• and
• encoding the current block based on the initial motion vectors of the current block and motion vector corrections, to obtain a bitstream;
• wherein the remaining DMVR process steps comprises:
• performing integer distance refinement initially at integer distance from the initial motion vectors, to obtain cost function values at integer distance positions, wherein a position with a lowest cost function value corresponds to a final integer distance;
• obtaining sub-pixel distance based on the cost function values at the integer distance positions and the cost function value at the position with the lowest cost function value;
• and
• obtaining the motion vector corrections based on the final integer distance and the sub-pixel distance.

A non-transitory computer-readable medium storing a bitstream generated by performing the following operations:

• obtaining initial motion vectors of a current block of a video frame, wherein the initial motion vectors comprise a first motion vector and a second motion vector;
• obtaining interpolated samples corresponding to two reference blocks of the current coding block based on the initial motion vectors from a bilinear interpolation, wherein the two reference blocks correspond to the first motion vector and the second motion vector, respectively;
• computing a sum of absolute differences (SAD) between the two reference blocks using a subset of the interpolated samples;
• determining the SAD is greater than or equal to a threshold value, wherein the threshold value is determined based on a size of the current block;
• performing remaining DMVR process steps;
• and
• encoding the current block based on the initial motion vectors of the current block and motion vector corrections, to obtain the bitstream;
• wherein the remaining DMVR process steps comprises:
• performing integer distance refinement initially at integer distance from the initial motion vectors, to obtain cost function values at integer distance positions, wherein a position with a lowest cost function value corresponds to a final integer distance;
• obtaining sub-pixel distance based on the cost function values at the integer distance positions and the cost function value at the position with the lowest cost function value;
• and
• obtaining the motion vector corrections based on the final integer distance and the sub-pixel distance.