Inter prediction method and related apparatus

Patent metadata

Field	Value
Publication number	US-11601667-B2
Application number	US-202117337556-A
Country	US
Kind code	B2
Filing date	Jun 3, 2021
Priority date	Dec 3, 2018
Publication date	Mar 7, 2023
Grant date	Mar 7, 2023

Abstract

This application provides an example inter prediction method and a related example apparatus. One example method includes parsing a bitstream to determine prediction information of a to-be-processed picture block, where the prediction information indicates that a prediction direction of the to-be-processed picture block is bidirectional prediction, where the bitstream does not include target identification information, and where the target identification information indicates to perform local illumination compensation (LIC) on the to-be-processed picture block; and obtaining a prediction value of the to-be-processed picture block based on the prediction information.

First claim

What is claimed is:

1. An inter prediction method, wherein the method comprises:
parsing a bitstream to determine prediction information of a to-be-processed picture block, wherein when the prediction information indicates that a prediction direction of the to-be-processed picture block is bidirectional prediction, the bitstream does not comprise target identification information, and wherein the target identification information indicates to perform local illumination compensation (LIC) on the to-be-processed picture block;
obtaining a prediction value of the to-be-processed picture block based on the prediction information;
parsing a second bitstream to determine second prediction information of a second to-be-processed picture block, wherein when the second prediction information indicates that a prediction direction of the second to-be-processed picture block is unidirectional prediction, the second bitstream comprises second target identification information, and wherein the second target identification information indicates to perform the LIC on the second to-be-processed picture block; and
obtaining a prediction value of the second to-be-processed picture block based on the second prediction information.

2. The method according to claim 1 , wherein in response to the prediction information of the to-be-processed picture block indicating that a prediction mode of the to-be-processed picture block is a merge mode, before obtaining the prediction value of the to-be-processed picture block based on the prediction information, the method further comprises:
determining at least one first information group, wherein each first information group comprises a motion vector, a prediction direction, and an LIC identifier; and
in response to a quantity of the at least one first information group being less than a target preset value, determining at least one second information group based on the at least one first information group, wherein each second information group comprises a motion vector, a prediction direction, and an LIC identifier, and wherein
obtaining the prediction value of the to-be-processed picture block based on the prediction information comprises:
obtaining the prediction value of the to-be-processed picture block based on the prediction information, the at least one first information group, and the at least one second information group.

3. The method according to claim 2 , wherein the at least one first information group comprises a first selection information group and a second selection information group, the at least one second information group comprises a first synthetic information group, and determining the at least one second information group based on the at least one first information group comprises:
using an average value of a motion vector in the first selection information group and a motion vector in the second selection information group as a motion vector in the first synthetic information group.

4. The method according to claim 2 , wherein the prediction direction comprises either the bidirectional prediction or the unidirectional prediction, the unidirectional prediction comprises first-directional prediction and second-directional prediction, the at least one first information group comprises a first selection information group and a second selection information group, the at least one second information group comprises a first synthetic information group, and determining the at least one second information group based on the at least one first information group comprises:
determining a prediction direction in the first synthetic information group based on a prediction direction in the first selection information group and a prediction direction in the second selection information group.

5. The method according to claim 4 , wherein in response to the LIC being performed on the to-be-processed picture block, an LIC identifier of the to-be-processed picture block is a first value, or in response to the LIC not being performed on the to-be-processed picture block, an LIC identifier of the to-be-processed picture block is a second value, and wherein
determining the at least one second information group based on the at least one first information group comprises one of:
in response to the prediction direction in the first synthetic information group being the bidirectional prediction, determining that an LIC identifier in the first synthetic information group is the second value;
in response to the prediction direction in the first synthetic information group being the unidirectional prediction, and an LIC identifier in the first selection information group and an LIC identifier in the second selection information group each being the second value, determining that an LIC identifier in the first synthetic information group is the second value; or
in response to the prediction direction in the first synthetic information group being the unidirectional prediction, and at least one of an LIC identifier in the first selection information group or an LIC identifier in the second selection information group being the first value, determining that an LIC identifier in the first synthetic information group is the first value.

6. The method according to claim 1 , wherein in response to the prediction information of the to-be-processed picture block indicating that a prediction mode of the to-be-processed picture block is a merge mode, before obtaining the prediction value of the to-be-processed picture block based on the prediction information, the method further comprises:
determining at least one third information group, wherein each third information group comprises a motion vector and an LIC identifier; and
in response to a quantity of the at least one third information group being less than a target preset value, determining at least one fourth information group based on the at least one third information group, wherein each fourth information group comprises a motion vector and an LIC identifier, and wherein
obtaining the prediction value of the to-be-processed picture block based on the prediction information comprises:
obtaining the prediction value of the to-be-processed picture block based on the prediction information, the at least one third information group, and the at least one fourth information group.

7. The method according to claim 6 , wherein the prediction direction comprises either the bidirectional prediction or the unidirectional prediction, the unidirectional prediction comprises first-directional prediction and second-directional prediction, the at least one third information group comprises a third selection information group and a fourth selection information group, the at least one fourth information group comprises a second synthetic information group, and determining the at least one fourth information group based on the at least one third information group comprises:
using a motion vector that is of the first-directional prediction and that corresponds to the third selection information group as a motion vector that is of the first-directional prediction and that corresponds to the second synthetic information group, and using a motion vector that is of the second-directional prediction and that corresponds to the fourth selection information group as a motion vector that is of the second-directional prediction and that corresponds to the second synthetic information group; or
using a motion vector that is of the second-directional prediction and that corresponds to the third selection information group as a motion vector that is of the first-directional prediction and that corresponds to the second synthetic information group, and using a motion vector that is of the first-directional prediction and that corresponds

Assignees

• Huawei Tech Co Ltd

Inventors

• Yang Haitao
• Mu Fan
• Zhang Lian
• Ma Xiang

Classifications

• H04N21/2343

  involving reformatting operations of video signals for distribution or compliance with end-user requests or end-user device requirements {(video transcoding H04N19/40; media packet handling at the source H04L65/762)} · CPC title

• H04N19/463

  by compressing encoding parameters before transmission · CPC title

• H04N19/52 Primary

  by predictive encoding · CPC title

• H04N19/107 Primary

  between spatial and temporal predictive coding, e.g. picture refresh · CPC title

• H04N19/82

  involving filtering within a prediction loop · CPC title