Transform Skip Mode for Large Video Blocks

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current video coding standards face challenges in efficiently encoding and decoding large video blocks, particularly in supporting diverse transform types and reducing computational complexity, especially for screen contents and natural video.

Innovation Solution

The proposed solution involves enabling the Transform Skip (TS) mode for larger block sizes, avoiding transform processes for large blocks, splitting residual blocks into regions for transform units, and applying sub-block transforms with identity or discrete transforms, along with QR-BDPCM for efficient coding.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If transform operations are applied to all video blocks, then coding precision is improved, but computational complexity increases

Engineering Contradiction:
Improvecoding precisionVSAvoidcomputational complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent applies transform operations selectively based on block characteristics rather than uniformly to all blocks. Small blocks (4x4, 8x8) receive full transform processing for high precision, while large blocks (16x16 and above) skip transform operations to reduce complexity, achieving local optimization of quality versus complexity trade-off

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent changes the processing parameters dynamically based on block size. A threshold-based parameter change mechanism enables or disables transform operations according to block dimensions, allowing the system to adapt computational intensity to the specific requirements of each block

Inventive Principle:
Principle #35Parameter changes

2Productivity

If larger video blocks are processed with multiple transform types, then coding efficiency is improved, but processing time increases

Engineering Contradiction:
Improvecoding efficiencyVSAvoidprocessing time
Core Design Contradiction:
ProductivityVSLoss of time

Solution Approach 1:

The patent applies different processing quality levels to different block sizes. Small blocks receive comprehensive multi-transform processing for maximum efficiency, while large blocks use simplified processing (skipping transforms) to reduce processing time, achieving local optimization of the efficiency-time trade-off

Inventive Principle:
Principle #3Local quality

3Device complexity

If transform operations are skipped for large blocks, then computational complexity is reduced, but coding precision deteriorates

Engineering Contradiction:
Improvecomputational complexityVSAvoidcoding precision
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The patent segments the video picture into blocks of different sizes and applies appropriate processing to each segment. By dividing the large picture into multiple smaller blocks, the system can apply transform operations to smaller segments where precision matters most, while skipping transforms for larger segments where complexity reduction is prioritized

Inventive Principle:
Principle #1Segmentation

4Productivity

If sub-block transforms are applied to large blocks, then coding efficiency is improved, but device complexity increases

Engineering Contradiction:
Improvecoding efficiencyVSAvoiddevice complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent segments large blocks into smaller sub-blocks and applies transforms to these sub-blocks. This segmentation approach allows the system to maintain coding efficiency through transform processing while managing device complexity by limiting transform operations to smaller sub-blocks rather than processing entire large blocks

Inventive Principle:
Principle #1Segmentation

Data Source

PatentUS11870996B2Transform bypass coded residual blocks in digital video
Publication Date: 2024.01.09 BYTEDANCE INC
  • US11870996B2 patent drawing
  • US11870996B2 patent drawing
  • US11870996B2 patent drawing

AI summary

A method for video processing includes determining that a size of a residual block of a video is larger than a maximum block size allowed for a transform skip mode, in which, during a decoding operation, residual coefficients are decoded without performing an inverse transform operation; splitting, based on the size of a residual block, the residual block into multiple regions, wherein each region represents a portion of the residual block of a difference between a portion of a current video block and a prediction block corresponding to the portion of the current video block; and determining a reconstructed current video block from the residual block based on selectively performing the inverse transform operation according to an indicator that indicates the transform skip mode.