Coefficient Group Context Modeling for Modular Video Coding

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current video encoding standards, such as H.264/AVC and developing MPEG-H, face inefficiencies in encoding and decoding residual video data due to the high percentage of data occupied by quantized transform coefficients, particularly in the encoding of significant-coefficient flags and coefficient levels, which hampers compression efficiency.

Innovation Solution

The implementation of multi-level significance maps and context derivation methods for determining context when encoding and decoding significant-coefficient flags and coefficient levels, allowing for more efficient processing by modifying the context neighborhood to reduce overhead and improve modular processing.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If extensive context modeling is used for significant-coefficient flags, then encoding accuracy is improved, but decoding complexity and computational overhead increase

Engineering Contradiction:
Improveencoding accuracyVSAvoiddecoding complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The transform unit is divided into multiple coefficient groups (e.g., 4x4 or 8x8 blocks), and context modeling is applied segmentarily to each group rather than to the entire transform unit. This segmentation allows the use of a limited, fixed context neighborhood within each group, reducing decoding complexity while maintaining adequate encoding accuracy through localized context adaptation.

Inventive Principle:
Principle #1Segmentation

2Measurement precision

If extensive context neighborhood is used for significant-coefficient flags, then encoding accuracy is improved, but memory requirements increase

Engineering Contradiction:
Improveencoding accuracyVSAvoidmemory requirements
Core Design Contradiction:
Measurement precisionVSQuantity of substance

Solution Approach 1:

A fixed, limited context neighborhood is defined for each coefficient group, using only nearby significant-coefficient flags within the same group (e.g., flags in the right column, bottom row, and diagonally adjacent positions). This local quality approach ensures that memory requirements are bounded and manageable, while still providing sufficient contextual information for accurate encoding within each localized region.

Inventive Principle:
Principle #3Local quality

3Ease of manufacture

If multi-level significance maps are implemented, then modular processing is enhanced, but context determination overhead increases

Engineering Contradiction:
Improvemodular processingVSAvoidcontext determination overhead
Core Design Contradiction:
Ease of manufactureVSLoss of time

Solution Approach 1:

The context neighborhood is pre-defined and fixed for each coefficient group, specifying exactly which nearby significant-coefficient flags should be used for context determination. This preliminary action eliminates the need for dynamic context neighborhood selection during decoding, reducing context determination overhead while maintaining the modular processing benefits of multi-level significance maps.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP2618573A1Methods and devices for context modeling to enable modular processing
Publication Date: 2013.07.24 BLACKBERRY LTD
  • EP2618573A1 patent drawingFigure 1
  • EP2618573A1 patent drawingFigure 2
  • EP2618573A1 patent drawingFigure 3

AI summary

Methods of encoding and decoding for video data are described for encoding or decoding coefficients for a transform unit. In particular, the significant-coefficient flags for a coefficient group are encoded and decoded based upon a context determination, and the context is determined based upon the values of neighboring flags. The neighborhood used to determine the context varies depending on whether the significant-coefficient flag to be encoded or decoded is in the right column or bottom row of the coefficient group or not. If it is in the right column or bottom row one of the alternative context neighborhoods is used to avoid relying on significant-coefficient flags in other coefficient groups except for the flags immediately adj acent the right border and bottom border of the coefficient group, and the flag diagonally to the lower-right.