Decoder-Side Motion Vector Refinement Subsampling

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing video coding technologies face challenges in efficiently refining motion vectors, leading to high computational complexity and distortion in bilateral matching processes.

Innovation Solution

The proposed solution involves subsampling motion vectors at a subblock level using processed initial predictors, reducing distortion computation complexity through techniques like bilateral matching and decoder-side motion vector refinement.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If bilateral matching distortion computation is performed using original predictors at subblock or whole block level, then motion vector refinement accuracy is improved, but computational complexity increases

Engineering Contradiction:

Improvemotion vector refinement accuracyVSAvoidcomputational complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent divides the block into multiple subblocks and performs bilateral matching distortion computation separately for each subblock using subsampled predictors. This segmentation allows the system to maintain refinement accuracy at the subblock level while reducing the overall computational complexity by processing smaller regions independently with subsampled data.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent applies subsampling to the predictors used in bilateral matching distortion computation, using only a subset of the full predictor data. This partial action approach reduces the computational load by processing fewer samples while still maintaining adequate motion vector refinement accuracy through the selective use of subsampled predictor information.

Inventive Principle:

Principle #16Partial or excessive action

2Device complexity

If subsampled subblock is used for deriving subblock motion vectors, then computational complexity is reduced, but measurement precision may deteriorate

Engineering Contradiction:

Improvedistortion computation complexityVSAvoidmotion vector accuracy

Core Design Contradiction:

Device complexityVSMeasurement precision

Solution Approach 1:

The patent segments the block into multiple subblocks and performs motion vector derivation for each subblock independently using subsampled predictors. This segmentation enables the use of reduced-complexity subsampled data at the subblock level while maintaining overall motion vector accuracy through localized processing that captures subblock-specific motion characteristics.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent applies different processing approaches to different parts of the block by performing subblock-level bilateral matching with subsampled predictors. This local quality approach allows the system to use subsampled data locally at each subblock where it is sufficient, while maintaining higher accuracy where needed through the refined motion vector derivation process applied to each subblock individually.

Inventive Principle:

Principle #3Local quality

Data Source

PatentUS20250211754A1Systems and methods for decoder-side motion vector refinement subsampling

Publication Date: 2025.06.26 TENCENT AMERICA LLC

AI summary

An example method of video coding includes receiving a video bitstream comprising a plurality of blocks, including a current block. The method also includes identifying a current subblock in the current block and deriving a subsampled subblock by sub-sampling the current subblock. The method further includes deriving, using the subsampled subblock, a set of subblock motion vectors for the current subblock, and deriving a set of refined subblock motion vectors from the set of subblock motion vectors. Additionally, the method includes reconstructing the current subblock using the derived set of refined subblock motion vectors.