Video Residual Coefficient Decoding Using Hierarchical Significance Maps

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The complexity of high efficiency video coding (HEVC) standards, particularly in encoding and decoding residual coefficients, is exacerbated by the need for numerous look-up tables, which consumes memory and increases complexity, especially when handling transform units of varying shapes and sizes.

Innovation Solution

The proposed solution involves a method for encoding and decoding residual coefficients using a two-layer hierarchy scan pattern, specifically a backward diagonal down-left scan, which reduces memory usage by dividing transform units into sub-blocks and using smaller scan patterns, such as 2x2 and 4x4, to encode significant coefficient groups, thereby minimizing the number of required look-up tables.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If transform units are subdivided into sub-blocks with multiple scan patterns, then coding efficiency is improved, but device complexity and memory consumption increase due to numerous look-up tables

Engineering Contradiction:
Improvecoding efficiencyVSAvoiddevice complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The transform unit is segmented into multiple sub-blocks, and the significance map is divided into corresponding sub-regions. Each sub-block has its own simplified scan pattern and look-up table, reducing the complexity of managing large transform units while maintaining coding efficiency through localized processing.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces a hierarchical dimension by organizing scan patterns and look-up tables at multiple levels (transform unit level and sub-block level). This dimensional organization allows efficient access to scan patterns based on transform unit size and shape without requiring all possible scan patterns to be stored simultaneously.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Productivity

If transform units are subdivided into sub-blocks with multiple scan patterns, then coding efficiency is improved, but memory consumption increases due to numerous look-up tables

Engineering Contradiction:
Improvecoding efficiencyVSAvoidmemory consumption
Core Design Contradiction:
ProductivityVSQuantity of substance

Solution Approach 1:

The large transform unit is segmented into smaller sub-blocks, each with its own simplified look-up table. This segmentation reduces the total memory required compared to storing comprehensive look-up tables for all possible transform unit configurations, while still providing the flexibility needed for efficient coding.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent implements a nested structure where sub-block look-up tables are contained within the broader transform unit processing framework. Smaller scan patterns for sub-blocks are nested within the larger transform unit scan patterns, allowing efficient memory utilization by storing only the necessary scan patterns at each hierarchical level.

Inventive Principle:
Principle #7Nested doll (Nesting)

3Adaptability or versatility

If the coding standard supports transform units of varying shapes and sizes, then adaptability is improved, but device complexity increases due to numerous look-up tables

Engineering Contradiction:
ImproveadaptabilityVSAvoiddevice complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The transform unit is segmented into sub-blocks that can independently handle different shapes and sizes. This segmentation allows the system to adapt to various transform unit configurations by applying appropriate scan patterns to sub-blocks without requiring complex overall processing logic for each possible transform unit size and shape combination.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent implements dynamic selection of scan patterns and processing methods based on the actual transform unit size and shape. The system can adaptively choose between different scan patterns (e.g., diagonal, horizontal, vertical) and processing approaches depending on the specific configuration, providing flexibility without requiring all possible configurations to be handled by complex fixed logic.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentEP3972249B1Decoding of the significance map for residual coefficients of a transform unit
Publication Date: 2023.04.05 CANON KK
  • EP3972249B1 patent drawingFigure 1
  • EP3972249B1 patent drawingFigure 2
  • EP3972249B1 patent drawingFigure 3A

AI summary

Disclosed is a method of decoding residual coefficients of a transform unit from a bitstream of video data. The method receives the transform unit (1500) from the bitstream of video data in which the transform unit has upper (1503) and lower (1502) level square layers. The upper level layer represents a square arrangement of at most four significant coefficient group flags with each of the significant coefficient group flags representing a non-overlapping region of the lower level layer. The method determines determining the significant coefficient group flags of the square upper level layer for the received transform unit, and determines values of residual coefficients of the square lower layer according to the determined significant coefficient group flags to decode the transform unit of the bitstream of video data. Corresponding encoding methods are also disclosed.