Image Decoding Hierarchical Partition Information Optimization

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional image coding methods, such as H.26x and MPEG-x, face inefficiencies in transmitting partition information due to the increased number of block-size types, leading to inefficient encoding and decoding of partitioned shapes in video signals.

Innovation Solution

An image coding method that determines a partitioning pattern hierarchically, generates partition information including maximum and minimum used hierarchy depths, and optionally omits information for unpartitioned blocks, to reduce coding overhead, and uses predicted partitioning patterns to further minimize coding requirements.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If the number of block-size types is increased to support various partitioning patterns, then the adaptability of the coding system is improved, but the complexity of transmitting partition information increases

Engineering Contradiction:
Improvepartitioning pattern flexibilityVSAvoidpartition information transmission complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The partition information transmission is segmented into two hierarchical levels: macroblock type information (coarse level) and sub-macroblock type information (fine level). This segmentation allows the system to first establish broad partitioning categories and then provide detailed specifications only where needed, reducing overall transmission complexity while maintaining support for multiple block-size types

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces a hierarchical dimension to partition information transmission by organizing block-size types into nested levels (macroblock → sub-macroblock). This dimensional organization transforms the flat complexity of transmitting all possible block-size combinations into a structured hierarchy, where information is transmitted in stages from general to specific

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

2Manufacturing precision

If hierarchical partitioning with multiple block-size types is implemented, then the coding precision is improved, but the amount of partition information to be transmitted increases

Engineering Contradiction:
Improvepartition information precisionVSAvoidpartition information volume
Core Design Contradiction:
Manufacturing precisionVSQuantity of substance

Solution Approach 1:

The patent applies partial action by transmitting sub-macroblock type information selectively rather than universally. Sub-macroblock partitioning information is transmitted only for macroblocks that require further partitioning, avoiding the transmission of redundant information for macroblocks that are already adequately represented at the coarser level, thus maintaining precision while reducing information volume

Inventive Principle:
Principle #16Partial or excessive action

3Measurement precision

If detailed sub-macroblock partition information is transmitted for all macroblocks, then the decoding accuracy is improved, but the coding overhead increases

Engineering Contradiction:
Improvepartition decoding accuracyVSAvoidcoding overhead
Core Design Contradiction:
Measurement precisionVSLoss of information

Solution Approach 1:

The patent applies local quality by providing different levels of partition information detail for different macroblocks based on their specific characteristics. Some macroblocks receive detailed sub-macroblock partition information while others use only the coarser macroblock type information, optimizing the balance between decoding accuracy and coding overhead for each local region

Inventive Principle:
Principle #3Local quality

Data Source

PatentEP3389277B1Image decoding method, and image decoding device
Publication Date: 2020.09.23 SUN PATENT TRUST
  • EP3389277B1 patent drawingFigure 1
  • EP3389277B1 patent drawingFigure 2A~2C
  • EP3389277B1 patent drawingFigure 3A~3C

AI summary

An image coding method according to the present invention partitions as input image signal (120) into processing units, and codes the partitioned image to generate a code sequence (140), the image coding method including: determining a partitioning pattern for hierarchically partitioning the input image signal (120) in order starting from a largest unit of the processing units; generating partition information (131) indicative of the partitioning pattern; and coding partition information (131), wherein the partition information (131) includes maximum used hierarchy depth information (used_max_depth) indicative of a maximum used hierarchy depth which is a hierarchy depth of a deepest processing unit of the processing units included in the partitioning pattern.