Video Coding Partition Depth Adjustment for Computational Complexity Reduction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current video coding technologies, such as those in the HEVC standard, face high computational complexity when encoding image components like depth and texture images, as they require extensive partitioning operations on both components to achieve optimal coding performance, leading to increased processing costs.

Innovation Solution

A method and device for coding a current block of a first image component relative to a reference block of a second image component, where the partitioning depth of the current block is adjusted based on the partitioning depth of the reference block, allowing for reduced computational complexity by either stopping partitioning at a sufficient depth or initializing from a depth close to the reference block's depth, depending on the image component types.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If extensive partitioning operations are performed on both depth and texture images to achieve optimal coding performance, then coding performance is improved, but computational complexity increases

Engineering Contradiction:
Improvecoding performanceVSAvoidcomputational complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent applies the copying principle by using the partitioning structure of the reference image component (depth or texture) as a template for encoding the current image component. Instead of performing independent extensive partitioning on both components, the invention copies the partitioning depth and structure from the reference block to the current block, significantly reducing computational complexity while maintaining coding performance through the similarity between corresponding blocks of the same scene

Inventive Principle:
Principle #26Copying

Solution Approach 2:

The patent applies preliminary action by performing partitioning on one image component first (the reference component), then using the resulting partitioning structure to guide the encoding of the second image component. This preliminary partitioning action eliminates the need to perform redundant partitioning operations on both components, reducing overall computational complexity while preserving coding performance

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP2868088B1Method for encoding a current block of a first image component relative to a reference block of at least one second image component, encoding device and corresponding computer program
Publication Date: 2018.10.31 ORANGE SA
  • EP2868088B1 patent drawingFigure 1A
  • EP2868088B1 patent drawingFigure 1B
  • EP2868088B1 patent drawingFigure 2A

AI summary

The invention concerns the encoding of at least one current block (CTBu; CTB'u) of a first image component relative to a reference block of at least one second image component (Cl21), said first and second image components being representative of a same scene, said reference block having previously been subjected to encoding by partitioning, then to decoding, said partitioning of the reference block being carried out a plurality of times until a predefined level (k) (k>0) of depth of partitioning is obtained. The method of encoding according to the invention comprises the steps which consist, on the basis of the type of the first and second image components, of: - either partitioning (C33a)) the current block a plurality of times until a level of depth of partitioning (k') dependent on the level of partitioning (k) of the reference block is obtained, - or partitioning (C34b)) the current block based on a level of depth of partitioning previously initialised (C33b)) by a level of depth of partitioning (k') that depends on the level of depth of partitioning (k) of the reference block.