Modified CABAC Transform Skip for HEVC High Bit-Rate Throughput

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The existing HEVC CABAC encoding and decoding methods face challenges in high bit-rate encoding, particularly in terms of throughput performance, due to computational complexity and resource consumption, especially when dealing with high bit-rate cases and higher bit-depth coding, which results in precision loss after scaling operations.

Innovation Solution

The proposed solution involves a high throughput binarization mode and modified transform skip processing, which includes adaptive binarization using VLC tables and a reduced Rice parameter update table, along with selective use of context models and coding modes to optimize encoding and decoding efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If standard CABAC encoding is used for high bit-rate cases, then encoding precision is maintained, but throughput performance deteriorates due to computational complexity

Engineering Contradiction:
Improveencoding precisionVSAvoidthroughput performance
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent segments the CABAC encoding process into different modes (bypass mode and regular mode) that can be selectively applied to different blocks or coefficient groups. This segmentation allows the system to use computationally intensive regular mode only where necessary for precision, while using simpler bypass mode for other regions, thereby resolving the contradiction between encoding precision and throughput performance.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies partial action by selectively enabling detailed CABAC processing only for specific blocks or coefficient regions that require high precision, rather than applying it uniformly across the entire image. This partial application of the complex encoding process maintains necessary precision while reducing overall computational complexity and improving throughput.

Inventive Principle:
Principle #16Partial or excessive action

2Productivity

If transform skip mode is used to reduce computational complexity, then throughput improves, but precision loss occurs in high bit-depth coding

Engineering Contradiction:
ImprovethroughputVSAvoidprecision in high bit-depth coding
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The patent implements local quality by allowing transform skip mode to be selectively applied to specific blocks or regions based on their characteristics, rather than uniformly across the entire image. This enables the system to maintain high precision in regions requiring it while using transform skip in regions where it is sufficient, thereby resolving the contradiction between throughput and precision in high bit-depth coding.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent introduces dynamic selection between transform mode and transform skip mode based on local image characteristics and bit-depth requirements. This dynamic adaptation allows the encoding process to automatically adjust its precision level in different regions, maintaining necessary precision where needed while improving throughput where possible.

Inventive Principle:
Principle #15Dynamics

3Productivity

If adaptive binarization with VLC tables is used, then encoding efficiency improves, but device complexity increases

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

Solution Approach 1:

The patent applies preliminary action by pre-computing and storing binarization tables (VLC tables) that can be directly looked up during encoding. This pre-computation shifts complexity from the real-time encoding process to an offline table generation process, allowing the encoder to achieve high encoding efficiency through simple table lookups rather than complex real-time calculations.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent uses copying by storing pre-computed binarization patterns in lookup tables that can be repeatedly referenced during encoding. Instead of recalculating binarization logic for each coefficient, the system copies pre-computed patterns from the tables, significantly reducing computational complexity while maintaining encoding efficiency.

Inventive Principle:
Principle #26Copying

Data Source

PatentEP3014879B1Modified coding for a transform skipped block for cabac in hevc
Publication Date: 2018.06.20 HUAWEI TECH CO LTD
  • EP3014879B1 patent drawingFigure 1
  • EP3014879B1 patent drawingFigure 2
  • EP3014879B1 patent drawingFigure 3

AI summary

In an example, an electronic device of a decoder is configured to obtain a bit stream and recover a binary symbol from the obtained bit stream. The electronic device is configured to determine whether the binary symbol is to be decoded using a modified transform skip mode. The electronic device is configured to, in response to determining that the binary symbol is not to be decoded using the modified transform skip mode, determine a first TS_Shift value, and recover video data using the first TS_Shift value. The electronic device is configured to, in response to determining that the binary symbol is to be decoded using the modified transform skip mode, determine a second TS_Shift value, and recover video data using the second TS_Shift value.