Video Coding Motion Vector Prediction Using Merge Modes

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing video coding technologies face challenges in efficiently reducing redundancy in video data, particularly in inter-picture prediction, where motion vectors (MVs) are not optimally predicted, leading to suboptimal compression ratios and increased data requirements.

Innovation Solution

The proposed solution introduces enhanced MV prediction mechanisms, including MMVD, sub-block affine merge prediction, and multi-hypothesis intra prediction, to improve the efficiency of motion vector coding by deriving MVs from refined merge modes and spatial or temporal neighbors, reducing the need for direct MV coding.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If motion vectors are coded directly without prediction, then coding simplicity is maintained, but compression efficiency deteriorates due to high bitrate requirements

Engineering Contradiction:
Improvecompression efficiencyVSAvoidcoding complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent applies preliminary action by predicting motion vectors from neighboring blocks before actual coding. The merge mode and AMVP mechanism pre-calculate candidate motion vectors from spatially adjacent blocks, so that the current block's motion vector can be derived with minimal additional data, improving compression efficiency while managing complexity through structured prediction workflows

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent uses intermediary elements in the form of motion vector candidates and candidate lists. Instead of directly coding motion vectors, the system introduces intermediate prediction candidates from neighboring blocks (A0, A1, B0, B1, etc.) that serve as mediators to derive the final motion vector, thereby reducing the bitrate required for direct motion vector transmission

Inventive Principle:
Principle #24Intermediary (Mediator)

2Productivity

If motion vectors are predicted from neighboring blocks, then compression ratio is improved, but prediction accuracy may deteriorate in complex motion scenarios

Engineering Contradiction:
Improvecompression ratioVSAvoidmotion vector accuracy
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The patent applies segmentation by dividing the motion vector prediction into multiple candidate blocks (A0, A1, B0, B1, B2) surrounding the current block. Each neighboring block provides a separate prediction candidate, allowing the encoder to select the most accurate one or combine multiple candidates, thereby maintaining compression efficiency while improving motion vector accuracy in complex scenarios

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent uses composite prediction by combining motion vector information from multiple neighboring blocks. The AMVP mode and merge mode synthesize candidates from different spatial locations (above, below, left, right neighbors), creating a composite prediction that is more robust to complex motion patterns than single-block prediction alone

Inventive Principle:
Principle #40Composite materials

3Productivity

If multiple prediction modes are implemented, then compression efficiency is enhanced, but computational complexity increases

Engineering Contradiction:
Improvecompression efficiencyVSAvoidcomputational complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent applies dynamics by implementing multiple prediction modes (merge mode, AMVP mode, refined merge mode) that can be dynamically selected based on the characteristics of each block. The encoder adapts between different prediction strategies depending on motion complexity, neighboring block availability, and rate-distortion optimization results, enhancing compression efficiency while managing computational complexity through selective mode application

Inventive Principle:
Principle #15Dynamics

Data Source

PatentEP3891975B1Method and apparatus for video coding
Publication Date: 2025.04.02 TENCENT AMERICA LLC
  • EP3891975B1 patent drawingFigure 1
  • EP3891975B1 patent drawingFigure 2
  • EP3891975B1 patent drawingFigure 3

AI summary

Aspects of the disclosure provide methods and apparatuses for video coding. In some examples, an apparatus includes processing circuitry that obtains a first syntax element from a coded video bitstream, the first syntax element being associated with a plurality of modes for identifying and applying motion information to a current block, and a first value of the first syntax element indicating a subset of the plurality of modes for the current block. The processing circuitry further generates a predictor of the current block according to the subset of the plurality of modes indicated by the first value of the first syntax element, and generates reconstructed samples of the current block based on the predictor.