Intra Merge Prediction Mode for Video Coding Efficiency

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

High-Efficiency Video Coding (HEVC) faces challenges in achieving optimal coding efficiency due to limitations in prediction modes, particularly in areas with complex textures, where pure intra or inter prediction modes fail to provide accurate predictions, leading to increased signaling overhead and reduced coding gains.

Innovation Solution

The introduction of Intra Merge Prediction (IMP) mode, which combines intra-prediction and merge modes by using a simple average or weighted sum of predictions from selected intra-prediction and merge modes, allowing for a combined prediction that reduces signaling overhead and improves coding efficiency in complex texture areas.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of information

If pure intra prediction mode is used, then signaling overhead is reduced, but prediction accuracy deteriorates in areas with complex textures

Engineering Contradiction:
Improvesignaling overheadVSAvoidprediction accuracy
Core Design Contradiction:
Loss of informationVSMeasurement precision

Solution Approach 1:

The patent combines intra prediction and merge modes into a unified Intra Merge Prediction mode. The predictor is generated by averaging the intra prediction result and the merge mode result, allowing the system to leverage both spatial and temporal information while maintaining compact signaling through merge-indexed prediction.

Inventive Principle:
Principle #5Merging (Combining)

2Measurement precision

If pure inter prediction mode is used, then prediction accuracy is improved for temporal correlation, but signaling overhead increases and coding gain is reduced in complex texture areas

Engineering Contradiction:
Improveprediction accuracyVSAvoidsignaling overhead
Core Design Contradiction:
Measurement precisionVSLoss of information

Solution Approach 1:

The patent merges intra prediction and inter prediction (merge mode) into a single hybrid mode. By using the average of both prediction types, it captures both spatial and temporal correlations while keeping signaling overhead low through the use of merge index selection.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The Intra Merge Prediction mode serves multiple functions: it can operate as pure intra prediction when beneficial, pure inter prediction when appropriate, or a combination of both. This multi-functionality allows adaptive optimization for different content types without requiring separate coding paths.

Inventive Principle:
Principle #6Universality (Multi-functionality)

3Measurement precision

If combined prediction mode (IMP mode) is used, then prediction accuracy is improved in complex texture areas, but device complexity increases

Engineering Contradiction:
Improveprediction accuracyVSAvoidcoding complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent applies a simple averaging operation to combine intra and inter predictions rather than using complex optimization algorithms. This partial action approach provides sufficient improvement for complex textures while keeping the computational complexity manageable.

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentUS11172203B2Intra merge prediction
Publication Date: 2021.11.09 HFI INNOVATION INC
  • US11172203B2 patent drawing
  • US11172203B2 patent drawing
  • US11172203B2 patent drawing

AI summary

A combined prediction mode for encoding or decoding a pixel block of a video picture is provided. When it is determined that the combined prediction mode is used, a video codec generates an intra predictor for the current block based on a selected intra-prediction mode and a merge-indexed predictor for the current block based on a selected merge candidate from a merge candidates list. The video codec then generates a final predictor for the current block based on the intra predictor and the merge-indexed predictor. The final predictor is then used to encode or decode the current block.