Coding Unit Division Using Variance to Cut HEVC Complexity

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The operation load and complexity in the coding unit division process of video coding standards like HEVC are high due to the recursive division of coding units, leading to excessive computational requirements.

Innovation Solution

A method to determine coding unit division by assessing the gradient of the current coding unit and its sub-coding units, allowing for the identification of units that do not need further division, thereby reducing operational complexity and load.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If recursive division of coding units is performed to achieve higher compression performance, then compression efficiency is improved, but operation load and complexity increase

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

Solution Approach 1:

The patent changes the parameter used for division decision from gradient to variance. By calculating the variance of pixel values within a coding unit and comparing it against a threshold, the system can determine whether further division is necessary. This parameter change simplifies the decision-making process while maintaining effective compression performance, as variance directly measures the heterogeneity of pixel values which is the true indicator of whether subdivision will yield benefits.

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent extracts only the essential information needed for division decision by using variance calculation rather than full gradient analysis. Instead of performing complex gradient computations for every coding unit, the system extracts a single variance value that captures the necessary information about pixel value distribution, thereby reducing computational load while preserving the ability to make accurate division decisions.

Inventive Principle:
Principle #2Taking out (Extraction)

2Measurement precision

If gradient calculation is performed for every coding unit to determine division, then division accuracy is improved, but computational burden increases

Engineering Contradiction:
Improvedivision decision accuracyVSAvoidcomputational burden
Core Design Contradiction:
Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The patent replaces gradient calculation with variance calculation for determining coding unit division. Variance is computed as the average of squared deviations from the mean pixel value, which is computationally simpler than gradient operations that require multiple pixel comparisons and arithmetic operations. This parameter substitution maintains measurement precision for division decisions while significantly reducing the computational energy required.

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent uses a simple variance threshold comparison as a disposable, low-cost method to determine whether further coding unit division is worthwhile. Instead of investing computational resources in complex gradient analysis for every coding unit, the system applies a quick variance check that can be computed with minimal operations, discarding detailed analysis for units that clearly meet or fail the division criterion.

Inventive Principle:
Principle #27Cheap short-living objects (Disposable)

Data Source

PatentEP3726837B1Coding unit division determining method and device, computing device and readable storage medium
Publication Date: 2025.11.19 TENCENT TECHNOLOGY (SHENZHEN) CO LTD
  • EP3726837B1 patent drawingFigure 1a~1b
  • EP3726837B1 patent drawingFigure 2~3
  • EP3726837B1 patent drawingFigure 4

AI summary

The present application discloses a method and apparatus for determining division of a coding unit, a computing device, and a readable storage medium. The method includes: determining whether a size of a current coding unit is greater than a size of a smallest coding unit; obtaining a gradient of the current coding unit when the size of the current coding unit is greater than the size of the smallest coding unit; obtaining gradients of sub-coding units included in the current coding unit, when the gradient of the current coding unit is less than a first threshold; and determining, according to the gradients of the sub-coding units, whether to divide the current coding unit. Compared with the existing technology, whether to further divide the current coding unit whose size is greater than the size of the smallest coding unit needs to be determined according to the gradient of the current coding unit and the gradients of the sub-coding units included in the current coding unit, so that a coding unit that does not need to be divided can be determined, thereby reducing an operation amount caused by the coding unit division and reducing an operation complexity degree.