Video Decoder Context Modeling for Intra Prediction Efficiency

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current video coding technologies face challenges in efficiently representing and decoding intra prediction directions, particularly in reducing redundancy and improving compression efficiency, especially with the increasing number of possible directions in newer coding standards like H.265 and VVC.

Innovation Solution

The method involves determining a prediction mode for a current coding unit, including inter and intra prediction modes, and using context indices based on syntax elements like transform unit coded block flags to perform arithmetic decoding for joint Cb Cr residual flags, allowing for efficient representation of chroma components as a single transform block.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If the number of intra prediction directions is increased to improve coding precision, then the compression efficiency deteriorates due to more bits required to represent directions

Engineering Contradiction:
Improveintra prediction direction precisionVSAvoidbitrate consumption
Core Design Contradiction:
Measurement precisionVSLoss of information

Solution Approach 1:

The patent changes the parameter representation by introducing context models that adaptively select among different intra prediction direction categories. Instead of directly encoding the precise direction index, the system uses context-based probability models to represent likely directions with fewer bits while maintaining the ability to encode less likely directions with higher precision when needed.

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent implements dynamic context modeling where the context index for arithmetic decoding varies based on previously decoded neighboring block directions. This dynamic adaptation allows the system to efficiently encode sequences of similar directions (common in video content) while maintaining flexibility to represent direction changes, thereby improving compression without sacrificing precision.

Inventive Principle:
Principle #15Dynamics

2Loss of information

If context modeling is used to reduce bits for direction representation, then the device complexity increases due to additional context index determination

Engineering Contradiction:
Improvebitrate consumptionVSAvoiddecoding complexity
Core Design Contradiction:
Loss of informationVSDevice complexity

Solution Approach 1:

The patent segments the intra prediction direction space into multiple context categories based on directional groups. By dividing the 33 directions into manageable context groups and assigning separate context models to each group, the system reduces the complexity of maintaining a single large context model while still achieving efficient compression through targeted probability adaptation within each segment.

Inventive Principle:
Principle #1Segmentation

3Loss of information

If joint Cb Cr residual coding is implemented to improve compression efficiency, then the manufacturing precision of chroma components may deteriorate

Engineering Contradiction:
Improvecompression efficiencyVSAvoidchroma component accuracy
Core Design Contradiction:
Loss of informationVSManufacturing precision

Solution Approach 1:

The patent applies joint Cb Cr residual coding selectively based on context conditions rather than universally. By using context indices to determine when joint coding is appropriate versus when separate coding should be used, the system achieves compression efficiency gains in suitable scenarios while maintaining chroma accuracy in scenarios where joint coding might be detrimental, thus balancing compression and precision requirements.

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP3881530B1Method and apparatus for video coding
Publication Date: 2024.12.18 TENCENT AMERICA LLC
  • EP3881530B1 patent drawingFigure 1
  • EP3881530B1 patent drawingFigure 2
  • EP3881530B1 patent drawingFigure 3

AI summary

A method of video decoding can include determining a prediction mode of a current coding unit (CU), and determining values of a transform unit (TU) coded block flag (CBF) of a Cb transform block, denoted tu_cbf_cb, and a TU CBF of a Cr transform block, denoted tu_cbf_cr, determining a context index, denoted ctxIdx, based on the prediction mode of the current CU and the values of the tu_cbf_cb, and the tu_cbf_cr, and performing an arithmetic decoding process according to a context model indicated by the ctxIdx to determine a bin of a joint Cb Cr residual (JCCR) flag indicating whether residual samples for both Cb and Cr chroma components of the current CU are coded as a single transform block.