Video Coding Transform Size Syntax for Larger Blocks

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current video coding standards, such as H.264/AVC, are limited in indicating transform sizes beyond 8x8, which restricts the ability to efficiently encode and decode video data using larger transform blocks, leading to suboptimal compression efficiency.

Innovation Solution

The introduction of a transform size syntax element with a flag value allows for the indication and interpretation of transform sizes greater than 8x8, enabling the use of larger transform blocks by combining the prediction block size and flag value to signal multiple transform sizes, thereby supporting additional block sizes like 16x16 and 8x16.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If transform sizes are limited to 4x4 and 8x8 as per current standards, then device complexity is reduced and ease of operation is maintained, but video coding efficiency and compression performance deteriorate

Engineering Contradiction:
Improvevideo coding efficiencyVSAvoidtransform size indication complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The transform size indication is segmented into multiple syntax elements: a base transform size element and additional flag elements. This segmentation allows the system to indicate larger transform sizes (beyond 8x8) by combining multiple simple binary flags, rather than requiring a complex multi-valued syntax element. Each flag represents a specific transform size option, and their combination enables flexible indication of transform sizes while maintaining simple individual component structures.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent extends the transform size indication from a single-dimension approach (limited to 4x4 and 8x8) to a multi-dimensional approach by introducing additional syntax elements that operate in different dimensional spaces. The combination of base transform size indicators with additional flag elements creates a multi-dimensional signaling mechanism that can represent larger transform sizes without increasing the complexity of individual signaling components.

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

2Loss of energy

If larger transform blocks (greater than 8x8) are used, then energy compaction improves and bit rates are reduced, but the ability to indicate and interpret transform sizes deteriorates due to standard limitations

Engineering Contradiction:
Improveenergy compactionVSAvoidtransform size indication capability
Core Design Contradiction:
Loss of energyVSAdaptability or versatility

Solution Approach 1:

The patent merges multiple syntax elements to indicate transform sizes. The base transform size element is combined with additional flag elements to collectively represent larger transform sizes. This merging allows the system to convey information about transform sizes greater than 8x8 by combining the information from multiple simpler elements, thereby achieving adaptability for larger transforms without requiring a complete overhaul of the existing syntax structure.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The transform size indication mechanism is made dynamic by allowing the system to adaptively select and combine different syntax elements based on the actual transform size being used. When larger transform sizes are required, additional flag elements are activated and combined with the base element, creating a flexible and dynamic indication system that can accommodate varying transform sizes as needed for optimal energy compaction.

Inventive Principle:
Principle #15Dynamics

3Productivity

If transform sizes beyond 8x8 are supported, then compression performance is improved, but current syntax elements become insufficient and require extension

Engineering Contradiction:
Improvecompression performanceVSAvoidsyntax element complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent prepares the syntax structure in advance by defining base transform size elements and预留 (reserving) additional flag elements that can be activated when needed. This preliminary action allows the system to smoothly extend support to larger transform sizes without requiring fundamental changes to the existing syntax framework. The pre-defined structure with extensible flags enables future-proofing of the syntax elements for supporting larger transforms while maintaining current functionality.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP2345253B1VIDEO CODING USING TRANSFORMS BIGGER THAN 4x4 AND 8x8
Publication Date: 2019.12.04 QUALCOMM INC
  • EP2345253B1 patent drawingFigure 1
  • EP2345253B1 patent drawingFigure 2
  • EP2345253B1 patent drawingFigure 3

AI summary

In a video processing system, a method and system for applying transforms larger than 8x8 and non-rectangular transforms, and generating transform size syntax elements indicative of the transforms for video decoding are provided. The transform size syntax element may be generated by an encoder based on a prediction block size of a video block and the contents of the video block. Further, the transform size syntax element may be generated according to a set of rules to select from 4x4, 8x8, and larger transform sizes during an encoding process. A decoder may perform an inverse transform based on the transform size syntax element and the rules used by the encoder. The transform size syntax element may be transmitted to the decoder as part of the encoded video bitstream.