Picture Prediction Using Mirror Reference Block Positions

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing video compression technologies face challenges in reducing complexity while improving picture prediction accuracy, particularly in inter prediction processes where high search complexity is incurred due to the need for template matching in forward and backward reference pictures.

Innovation Solution

A method that determines the positions of forward and backward reference blocks based on initial motion information, using a matching cost criterion to identify optimal reference blocks, thereby avoiding the need for template matching and simplifying the prediction process.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If decoder-side motion vector refinement (DMVR) technology is used to refine motion information, then picture prediction accuracy is improved, but search complexity increases due to template matching in forward and backward reference pictures

Engineering Contradiction:

Improvepicture prediction accuracyVSAvoidsearch complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent extracts only the essential motion information from the candidate list and directly uses it to determine reference block positions, eliminating the unnecessary template matching step. This extraction approach keeps only the core functional element (motion information) while removing the complex computational process (template matching), thereby improving accuracy without increasing search complexity

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

Instead of using template matching to find motion information and then refining it (the conventional DMVR approach), the patent inverts the process by directly using motion information from the candidate list to determine reference block positions. This inversion eliminates the template matching step entirely while still achieving accurate prediction

Inventive Principle:

Principle #13The other way round (Inversion)

2Measurement precision

If template matching is performed in forward and backward reference pictures, then motion information accuracy is improved, but coding overhead increases

Engineering Contradiction:

Improvemotion information accuracyVSAvoidcoding overhead

Core Design Contradiction:

Measurement precisionVSLoss of information

Solution Approach 1:

The patent uses inexpensive motion information directly from the candidate list without performing expensive template matching operations. By treating the candidate motion information as sufficient for the task, it avoids the high computational cost of template matching while still achieving acceptable accuracy, effectively using a 'cheap' solution instead of an 'expensive' one

Inventive Principle:

Principle #27Cheap short-living objects (Disposable)

Data Source

PatentUS20250016358A1Picture prediction method and apparatus, and codec

Publication Date: 2025.01.09 HUAWEI TECH CO LTD

AI summary

A picture prediction method is provided, which includes: obtaining initial motion information of a current picture block; determining, based on a matching cost criterion, that positions of a pair of reference blocks are a position of a target forward reference block of the current picture block and a position of a target backward reference block of the current picture block, where positions of each pair of reference blocks include a position of a forward reference block and a position of a backward reference block; and for the positions of each pair of reference blocks, a first position offset and a second position offset are in a mirror relationship; and obtaining a predicted value of a pixel value of the current picture block based on a pixel value of the target forward reference block and a pixel value of the target backward reference block.