Method and apparatus for video encoding and decoding

Patent metadata

Field	Value
Publication number	US-11323739-B2
Application number	US-202016990837-A
Country	US
Kind code	B2
Filing date	Aug 11, 2020
Priority date	Jun 20, 2018
Publication date	May 3, 2022
Grant date	May 3, 2022

Abstract

A video coding method includes obtaining a current frame from a plurality of video frames, from which at least two or more video frames from have different resolutions; determining an initial motion vector (MV) corresponding to each block to be encoded in the current frame at a corresponding resolution; and determining a target MV resolution according to a resolution configuration of the current frame to represent a target resolution. The method also includes determining a target MV corresponding to each block in the current frame at the target resolution represented by the target MV resolution; obtaining a motion vector prediction (MVP) corresponding to each block in the current frame at the target resolution; and encoding the current frame according to a motion vector difference (MVD) between each target MV and the corresponding MVP such that the MV and the corresponding MVP are at a same resolution.

First claim

What is claimed is:

1. A video coding method for a computer device, comprising:
obtaining a current frame from a plurality of video frames to be encoded, at least two or more video frames from the plurality of video frames having different resolutions;
determining an initial motion vector (MV) corresponding to each block to be encoded in the current frame at a corresponding resolution;
determining a target MV resolution according to a resolution configuration of the current frame to represent a target resolution;
determining a target MV corresponding to each block to be encoded in the current frame at the target resolution represented by the target MV resolution;
obtaining a motion vector prediction (MVP) corresponding to each block to be encoded in the current frame at the target resolution; and
encoding the current frame according to a motion vector difference (MVD) between each target MV and the corresponding MVP such that the MV and the corresponding MVP are at a same resolution,
wherein obtaining the current frame comprises:
obtaining an original video frame;
obtaining a processing manner corresponding to the original video frame; and
processing the original video frame in the processing manner, to obtain the current frame, the processing manner being a full resolution processing manner or a down-sampling processing manner.

2. The method according to claim 1 , wherein:
the current resolution is inconsistent with the target resolution represented by the target MV resolution; and
the determining a target MV corresponding to each block to be encoded in the current frame at a target resolution represented by the target MV resolution comprises:
transforming the initial MV to the target MV at the target resolution.

3. The method according to claim 2 , wherein the target resolution represented by the target MV resolution is an original resolution of the original video frame corresponding to the current frame; and
the transforming the initial MV to the target MV at the target resolution comprises:
transforming the initial MV corresponding to each block to be encoded of the current frame to the target MV at the original resolution when the current frame is obtained by processing the original video frame in the down-sampling processing manner.

4. The method according to claim 2 , wherein the transforming the initial MV to the target MV at the target resolution comprises:
determining a first vector transformation coefficient according to a current resolution of the current frame and the target MV resolution; and
transforming the initial MV to the target MV at the target resolution according to the first vector transformation coefficient.

5. The method according to claim 1 , wherein the obtaining an MVP corresponding to each block to be encoded and at the target resolution comprises:
determining, for a current block to be encoded in a process of encoding each block to be encoded in the current frame, a reference frame in which a reference block corresponding to the current block to be encoded is located;
determining, in the reference frame, an image block corresponding to the current block to be encoded with respect to an image location; and
obtaining a target MV corresponding to the image block and at the target resolution as an MVP of the current block to be encoded.

6. The method according to claim 5 , wherein the target resolution represented by the target MV resolution information is a current resolution of the current frame; and
the obtaining a target MV corresponding to the image block and at the target resolution as an MVP of the current block to be encoded comprises:
obtaining an initial MV of the image block at a resolution of the reference frame; and
transforming the initial MV of the image block to a target MV at the current resolution as the MVP of the current block to be encoded when the resolution of the reference frame is inconsistent with the current resolution.

7. The method according to claim 6 , wherein the transforming the initial MV of the image block to a target MV at the current resolution as the MVP of the current block to be encoded comprises:
obtaining a second vector transformation coefficient according to a resolution of the reference frame and the target MV resolution representing the current resolution; and
transforming, according to the second vector transformation coefficient, the initial MV of the image block to the target MV at the current resolution as the MVP of the current block to be encoded.

8. A video coding method for a computer device, comprising:
obtaining encoded data corresponding to a current frame;
extracting a motion vector difference (MVD) corresponding to each block to be decoded in the current frame in the encoded data, a resolution corresponding to the MVD being a target resolution corresponding to a resolution configuration, wherein the target resolution is a resolution of an original video frame corresponding to the encoded data or a resolution of the current frame;
determining a motion vector prediction (MVP) corresponding to each block to be decoded;
processing the MVD and the MVP corresponding to each block to be decoded at a same resolution after determining that the resolution of the current frame is inconsistent with the target resolution, to obtain a motion vector (MV) corresponding to the corresponding block to be decoded and at the resolution of the current frame;
determining a reference block corresponding to each block to be decoded according to an MV corresponding to each block to be decoded; and
obtaining a reconstructed video frame according to each reference block and the encoded data.

9. The method according to claim 8 , wherein the determining an MVP corresponding to each block to be decoded comprises:
determining, for a current block to be decoded in a process of processing each block to be decoded in the current frame, a reference frame in which a reference block corresponding to the current block to be decoded is located;
determining, in the reference frame, an image block corresponding to the current block to be decoded with respect to an image location; and
using an MV corresponding to the image block as an MVP of the current block to be decoded.

10. The method according to claim 9 , wherein the using an MV corresponding to the image block as an MVP of the current block to be decoded comprises:
obtaining an MV corresponding to the image block at a resolution of the reference frame; and
transforming the MV corresponding to the image block to an MV at the resolution of the current frame when the resolution of the reference frame is inconsistent with the resolution of the current frame, to obtain the MVP of the current block to be decoded.

11. The method according to claim 8 , wherein the processing the MVD and the MVP corresponding to each block to be decoded at the same resolution after determining that a resolution of the current frame is inconsistent with the target resolution, to obtain an MV corresponding to the corresponding block to be decoded and at the resolution of the current frame comprises:
transforming, for a current block to be decoded in a process of processing each block to be decoded in the current frame, an MVD corresponding to the current block to be decoded to a target MVD at the resolution of the current frame when an MVP corresponding to the current block to be decoded corresponds to the resolution of the current frame; and
obtaining an MV corresponding to the current block to be decoded and at the resolution of the current frame according to the target MVD and the MVP corresponding to the current block to be decoded.

12. The method a

Assignees

• Tencent Tech Shenzhen Co Ltd

Inventors

• Gao Xinwei
• Mao Xunan
• Gu Chenchen

Classifications

• H04N19/52 Primary

  by predictive encoding · CPC title

• H04N19/523

  with sub-pixel accuracy · CPC title

• H04N19/147

  according to rate distortion criteria (rate-distortion as a criterion for motion estimation H04N19/567) · CPC title

• H04N19/53

  Multi-resolution motion estimation; Hierarchical motion estimation · CPC title

• H04N19/172 Primary

  the region being a picture, frame or field · CPC title