MVD Scaling for Joint MVD Coding
AssigneeTencent America LLC
InventorsLiang ZHAO, Xin ZHAO, Shan LIU
Life of this patent
Estimating the term…
Abstract
This disclosure relates generally to video coding and particularly to methods and systems for derivation and scaling of motion vector difference (MVD) in joint MVD scaling for compound inter-prediction of video blocks, and the signaling therefor. In one example implementations, a method and video device for processing a video block of a video stream is disclosed. For example, the method may include scaling a jointly motion vector difference and quantize the scaled motion vector difference in order derive a predicted motion vector according to a pixel resolution of the predicted motion vector.
Claims20
A method for processing a video block in a video bitstream, comprising:
- decoding at least one syntax element from the video bitstream indicating that the video block is inter-predicted using a joint motion vector difference (MVD) coding mode, wherein a first MVD for a first reference frame and a second MVD for a second reference frame are jointly coded;
- decoding the first MVD from the video bitstream;
- scaling the first MVD using a scaling factor to generate the second MVD, wherein the scaling factor is predefined;
- and
- decoding the video block based on a first motion vector derived from the first MVD and a second motion vector derived from the second MVD.
A method for encoding a video block, comprising:
- determining that the video block is to be inter-predicted using a joint motion vector difference (MVD) coding mode, wherein a first MVD for a first reference frame and a second MVD for a second reference frame are jointly coded;
- encoding at least one syntax element in a video bitstream indicating the joint MVD coding mode;
- encoding the first MVD in the video bitstream;
- scaling the first MVD using a scaling factor to generate the second MVD, wherein the scaling factor is predefined;
- and
- encoding the video block based on a first motion vector derived from the first MVD and a second motion vector derived from the second MVD.
A method of encoding visual media data, the method comprising:
- encoding and transmitting a bitstream that includes the visual media data according to a format rule,
- wherein the bitstream includes:
- at least one syntax element indicating that a video block is inter-predicted using a joint motion vector difference (MVD) coding mode, wherein a first MVD for a first reference frame and a second MVD for a second reference frame are jointly coded;
- and
- the first MVD for the video block;
- wherein the format rule specifies that the second MVD is derived by scaling the first MVD using a scaling factor;
- wherein the scaling factor is predefined;
- and
- wherein the video block is decoded based on a first motion vector derived from the first MVD and a second motion vector derived from the second MVD.
Dependencies are inferred from the claim text.