Video Coding With Affine Motion and Optical-Flow Refinement

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing video coding technologies face challenges in efficiently compressing video data, particularly in reducing redundancy and optimizing motion vector prediction, which affects bandwidth and storage requirements.

Innovation Solution

The implementation of a video coding method that utilizes affine motion compensated prediction with optical flow refinement and motion vector clipping to enhance motion vector prediction accuracy and efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Quantity of substance

If motion vector prediction is used to reduce data required for coding motion vectors, then compression ratio is improved, but motion vector prediction accuracy may deteriorate

Engineering Contradiction:
Improvedata required for coding motion vectorsVSAvoidmotion vector prediction accuracy
Core Design Contradiction:
Quantity of substanceVSMeasurement precision

Solution Approach 1:

The current block is divided into multiple sub-blocks, and motion vectors are predicted separately for each sub-block using affine transformation. This segmentation allows more precise local motion representation while maintaining efficient coding through parameter-based prediction.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent uses affine transformation parameters (control point motion vectors) to describe motion across sub-blocks. By changing from individual motion vectors per block to parameter-based motion description, the patent reduces data quantity while maintaining prediction accuracy through mathematical transformation.

Inventive Principle:
Principle #35Parameter changes

2Measurement precision

If affine motion compensated prediction with optical flow refinement is used to improve motion vector prediction accuracy, then video quality is improved, but computational complexity increases

Engineering Contradiction:
Improvemotion vector prediction accuracyVSAvoidcomputational complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent performs motion compensation prediction using affine transformation before final reconstruction. By preparing control point motion vectors and applying affine transformation in advance, the patent improves prediction accuracy while managing computational complexity through structured preprocessing.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent introduces control point motion vectors as intermediaries between reference picture motion and current block motion. These control points serve as mediators that simplify the computational complexity by reducing the number of direct motion vector calculations needed while maintaining prediction accuracy.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Measurement precision

If motion vector clipping is applied to constrain motion vectors within target range, then prediction accuracy is improved, but loss of information increases

Engineering Contradiction:
Improveprediction accuracyVSAvoidmotion vector information
Core Design Contradiction:
Measurement precisionVSLoss of information

Solution Approach 1:

The patent applies motion vector clipping locally within target ranges specific to each current block's characteristics. By determining target ranges based on block size and position, the patent maintains prediction accuracy for local regions while preserving more motion vector information globally compared to uniform clipping.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent changes the state of motion vectors by clipping them to target ranges while maintaining the underlying affine transformation parameters. This parameter-based approach allows the system to constrain motion vectors for accurate prediction while retaining the ability to reconstruct original motion information when needed.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP3874756B1Method and apparatus for video coding
Publication Date: 2025.10.15 TENCENT AMERICA LLC
  • EP3874756B1 patent drawingFigure 1~2
  • EP3874756B1 patent drawingFigure 3~4
  • EP3874756B1 patent drawingFigure 5~6

AI summary

Aspects of the disclosure provide a method and an apparatus for video coding. In some examples, an apparatus includes processing circuitry that determines a plurality of first sub-block motion vectors for a plurality of sub-blocks of a current block, and determines a plurality of second sub-block motion vectors according to the plurality of first sub-block motion vectors and a first target range. The processing circuitry also determines a set of gradient values associated with a current sample location in a current sub-block, determines an adjustment vector associated with the current sample location, and determines a set of adjustment values associated with the current sample location according to the set of gradient values and the adjustment vector. The processing circuitry generates a predicted sample associated with the current sample location according to a combination of a corresponding sample in a reference sub-block and the set of adjustment values.