VVC Decoder-Derived Intra Prediction with Selective Angular Modes

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The computational complexity and performance of Decoder Side Intra Mode Derivation (DIMD) coding tool in video coding systems, particularly in VVC, are not optimally addressed, leading to inefficiencies in video decoding processes.

Innovation Solution

A method and apparatus for video coding that reduces computational complexity by determining a selected set of angular intra prediction candidate modes based on block size and neighboring block prediction modes, using statistics or histograms, and applying gradient filters to derive a final intra predictor, with optional signaling of delta angles to improve accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If all angular intra prediction modes are evaluated for DIMD, then prediction accuracy is improved, but computational complexity increases

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

Solution Approach 1:

The patent segments the full set of angular intra prediction modes into multiple subsets based on block characteristics (size, shape, neighboring block modes). Instead of evaluating all modes, the decoder selectively evaluates only relevant subsets, reducing computational complexity while maintaining prediction accuracy for different block types

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies partial action by evaluating only a subset of angular modes rather than all modes. The selection of which subset to evaluate depends on block characteristics, applying just enough computational effort to achieve good prediction accuracy without the excessive complexity of full mode evaluation

Inventive Principle:
Principle #16Partial or excessive action

2Productivity

If DIMD is applied to all block types, then coding efficiency is improved, but processing time increases

Engineering Contradiction:
Improvecoding efficiencyVSAvoidprocessing time
Core Design Contradiction:
ProductivityVSLoss of time

Solution Approach 1:

The patent applies local quality by adapting the DIMD process to different block types locally. Small blocks use a simplified mode subset evaluation, while larger blocks use more comprehensive evaluation. This ensures coding efficiency is improved where needed without uniformly increasing processing time for all blocks

Inventive Principle:
Principle #3Local quality

3Reliability

If the full set of angular intra prediction modes is used, then prediction performance is improved, but decoder complexity increases

Engineering Contradiction:
Improveprediction performanceVSAvoiddecoder complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent makes the decoder dynamic by adaptively selecting which angular mode subsets to evaluate based on real-time block characteristics (size, aspect ratio, neighboring block modes). This dynamic adaptation maintains high prediction performance when needed while reducing complexity for blocks where full evaluation is unnecessary

Inventive Principle:
Principle #15Dynamics

Data Source

PatentUS20250254292A1Method and Apparatus of Improvement for Decoder-Derived Intra Prediction in Video Coding System
Publication Date: 2025.08.07 MEDIATEK INC
  • US20250254292A1 patent drawing
  • US20250254292A1 patent drawing
  • US20250254292A1 patent drawing

AI summary

Methods and apparatus for video coding are disclosed. According to one method, a selected set of angular intra prediction candidate modes corresponding to a subset of an initial set of angular intra prediction modes is used to derive DIMD candidates. In another method, a delta angle between the final intra prediction mode and a DIMD derived mode is signalled or parsed. In yet another method, a DIMD candidate mode is determined by using a process including comparing gradient magnitudes of the gradient filtered results with a threshold.