Video Encoder CABAC Bypass Coding Residual Coefficients

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The existing video coding standards, such as H.264/MPEG-4 AVC, face challenges in achieving high compression efficiency when encoding residual coefficients, particularly at high resolutions or frame rates, due to limitations in the coding of residual data in the frequency domain.

Innovation Solution

The method involves a video encoder and decoder system that utilizes a transform unit (TU) with a quadtree structure, allowing for both square and non-square shapes, and employs context-adaptive binary arithmetic coding (CABAC) with bypass coding mode to efficiently encode and decode residual coefficients, enhancing throughput by reading multiple adjacent bins in parallel.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of information

If context-adaptive binary arithmetic coding (CABAC) is used to encode residual coefficients, then coding efficiency is improved, but processing complexity and time increase

Engineering Contradiction:
Improvecoding efficiencyVSAvoidprocessing complexity
Core Design Contradiction:
Loss of informationVSDevice complexity

Solution Approach 1:

The patent segments the encoding process by separating bypass-coded bins from context-modelled bins. Bypass-coded bins are processed independently and in parallel, while context-modelled bins use adaptive probability models. This segmentation allows the system to achieve high coding efficiency for bypass-coded regions without the full computational overhead of CABAC, thus resolving the contradiction between coding efficiency and processing complexity.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies partial CABAC by using bypass coding for certain bins that do not require context modeling. This partial application of the full CABAC process reduces processing complexity while maintaining sufficient coding efficiency for those specific bins, thereby resolving the contradiction between complete CABAC efficiency gains and the associated processing burden.

Inventive Principle:
Principle #16Partial or excessive action

2Productivity

If multiple adjacent bins are read in parallel to increase throughput, then processing speed is improved, but synchronization and control complexity increase

Engineering Contradiction:
ImprovethroughputVSAvoidsynchronization complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent segments bins into bypass-coded and context-modelled categories, allowing bypass-coded bins to be read in parallel without requiring complex synchronization with context-modelled bins. This segmentation enables throughput improvement through parallel processing while minimizing synchronization complexity by isolating the parallel operations to specific bin types.

Inventive Principle:
Principle #1Segmentation

3Adaptability or versatility

If transform units are allowed to have non-square shapes, then adaptability to different block structures is improved, but decoding complexity increases

Engineering Contradiction:
Improveblock structure adaptabilityVSAvoiddecoding complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent applies different coding strategies to different regions of the transform unit based on its shape. For non-square TUs, specific scanning orders and coefficient encoding methods are applied locally to match the geometric structure, while square TUs use standard procedures. This local adaptation improves versatility without requiring complete redesign of the decoding process for all TU types.

Inventive Principle:
Principle #3Local quality

Data Source

PatentEP3737094B1A method, apparatus and program for coding image data into a bitstream
Publication Date: 2022.10.12 CANON KK
  • EP3737094B1 patent drawingFigure 1
  • EP3737094B1 patent drawingFigure 2
  • EP3737094B1 patent drawingFigure 3A

AI summary

Disclosed is a method of coding image data into a bitstream, the method comprising: arithmetically coding flags each of which being for a different one of prediction units in a target coding unit into the bitstream, in a case where the target coding unit is divided into the prediction units, the flag indicating whether any of specific intra-prediction modes is used; bypass coding pieces of information each of which being for a different one of the prediction units in the target coding unit into the bitstream, each of the pieces of information being for designating one of intra-prediction modes; and coding the target coding unit into the bitstream using intra-prediction modes to be indicated by the flags and the pieces of information. In the bitstream, the pieces of information each of which being for a different one of the prediction units in the target coding unit are located posterior to the flags each of which being for a different one of the prediction units in the target coding unit.