Non-Dyadic Block Transforms for Efficient Video Coding

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing video coding standards face inefficiencies in handling non-dyadic blocks, leading to suboptimal bandwidth usage and compression performance.

Innovation Solution

Implementing transform coding techniques such as zero-out transforms, padding, and non-dyadic transforms for non-dyadic blocks, along with specific transform selection rules and secondary transforms like LFNST, to enhance coding efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If transform coding is applied to non-dyadic blocks using padding or splitting, then coding efficiency is improved, but device complexity increases

Engineering Contradiction:

Improvecoding efficiencyVSAvoidprocessing complexity

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

The patent divides non-dyadic blocks into multiple dyadic transform units (TUs) that can be processed using standard dyadic transforms. This segmentation allows the use of efficient, standardized transform operations on smaller regular blocks while handling the irregular non-dyadic block structure, thereby improving coding efficiency without requiring entirely new complex transform algorithms.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent introduces padding operations as an intermediary step that extends non-dyadic blocks to dyadic sizes, enabling the use of standard dyadic transform coding. This intermediary padding mechanism bridges the gap between irregular block structures and regular transform requirements, improving efficiency while maintaining compatibility with existing transform infrastructure.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Measurement precision

If non-dyadic transforms are used for non-dyadic blocks, then transform accuracy is improved, but device complexity increases

Engineering Contradiction:

Improvetransform accuracyVSAvoidtransform implementation complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

Instead of developing complex non-dyadic transforms to handle irregular blocks, the patent inverts the approach by transforming regular dyadic blocks to fit the non-dyadic structure through padding and splitting. This inversion allows the use of simpler, well-optimized dyadic transforms while achieving accurate representation of non-dyadic block content.

Inventive Principle:

Principle #13The other way round (Inversion)

3Loss of energy

If transform skip mode is applied to non-dyadic blocks, then bandwidth is reduced, but coding efficiency deteriorates

Engineering Contradiction:

Improvebandwidth consumptionVSAvoidcoding efficiency

Core Design Contradiction:

Loss of energyVSProductivity

Solution Approach 1:

The patent applies transform skip mode selectively based on local characteristics of non-dyadic blocks. By analyzing the content and complexity of each block, the encoder can decide whether to apply transform coding or skip it, allowing bandwidth reduction in suitable cases while maintaining coding efficiency where transforms are beneficial.

Inventive Principle:

Principle #3Local quality

Data Source

PatentUS12587644B2Transforms on non-dyadic blocks

Publication Date: 2026.03.24 DOUYIN VISION CO LTD

AI summary

A mechanism for processing video data is disclosed. A Unsymmetric Binary Tree (UBT) partition of a parent block is determined to create two sub-blocks with different dimensions. At least one of the sub-blocks includes a side that is non-dyadic. A conversion is performed between a visual media data and a bitstream based on the sub-blocks.