Video Decoder Prediction Mode List Adaptation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current video compression standards face challenges in efficiently coding prediction modes, leading to increased signaling costs and potential loss of precision due to the large number of available prediction modes, especially when certain modes provide similar predictions.

Innovation Solution

A method that adapts the list of prediction modes by determining a distance measurement between predictor blocks and modifying the list based on these measurements, optimizing the number and ordering of prediction modes to reduce signaling costs and improve prediction accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If a large number of intra prediction modes (35 modes in HEVC) are provided to predict a current block, then prediction accuracy is improved, but the signaling cost increases and coding complexity increases

Engineering Contradiction:
Improveprediction accuracyVSAvoidcoding complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments the 35 prediction modes into two distinct lists: an MPM list containing 3 most probable modes with lower signaling cost, and a non-MPM list containing the remaining 32 modes. This segmentation allows the decoder to first check the compact MPM list before considering the full non-MPM list, thereby reducing the average coding complexity while maintaining access to all 35 modes when needed for high prediction accuracy.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent dynamically adapts the prediction mode selection process by using context-based probability estimation. The MPM list is constructed dynamically based on the prediction modes of neighboring blocks, and the decoder dynamically determines whether to use the MPM list or non-MPM list based on the coded indication. This dynamic approach optimizes the balance between signaling cost and prediction accuracy for each specific block.

Inventive Principle:
Principle #15Dynamics

2Measurement precision

If a large number of intra prediction modes are provided to predict a current block, then prediction accuracy is improved, but the signaling cost increases

Engineering Contradiction:
Improveprediction accuracyVSAvoidsignaling cost
Core Design Contradiction:
Measurement precisionVSLoss of information

Solution Approach 1:

The patent segments the 35 prediction modes into two distinct lists: an MPM list containing 3 most probable modes with lower signaling cost, and a non-MPM list containing the remaining 32 modes. This segmentation allows the decoder to first check the compact MPM list before considering the full non-MPM list, thereby reducing the average coding complexity while maintaining access to all 35 modes when needed for high prediction accuracy.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent implements partial action by providing a simplified MPM list with only 3 modes for cases where the most probable modes are sufficient, while still maintaining the complete non-MPM list with 32 additional modes for cases where higher prediction accuracy is needed. The decoder selectively uses the appropriate list based on the coded indication, avoiding the overhead of signaling all 35 modes in every case while ensuring full capability when required.

Inventive Principle:
Principle #16Partial or excessive action

3Productivity

If certain prediction modes are removed to reduce signaling cost, then coding efficiency is improved, but prediction precision may be lost

Engineering Contradiction:
Improvecoding efficiencyVSAvoidprediction precision
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The patent dynamically adapts the prediction mode selection process by using context-based probability estimation. The MPM list is constructed dynamically based on the prediction modes of neighboring blocks, and the decoder dynamically determines whether to use the MPM list or non-MPM list based on the coded indication. This dynamic approach optimizes the balance between signaling cost and prediction accuracy for each specific block.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The patent changes the parameter of prediction mode availability by providing two different configurations: a compact MPM list with 3 modes for efficiency, and a complete non-MPM list with 32 additional modes for precision. The system selectively activates the appropriate configuration based on the coded indication, thereby adapting the prediction precision to match the coding efficiency requirements of each block.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP3456045B1Methods and devices for coding and decoding a data stream representative of at least one image
Publication Date: 2021.06.02 B COM
  • EP3456045B1 patent drawingFigure 1~2
  • EP3456045B1 patent drawingFigure 3A~3B
  • EP3456045B1 patent drawingFigure 4A~6

AI summary

The invention relates to a method and a device for decoding a coded data stream representative of at least one image cut up into blocks, at least two modes of prediction being available for predicting a block, termed the current block, of the image. For the said current block, and for each mode of prediction available for predicting the said current block, a predictor block associated with the available mode of prediction is obtained. Next, a list comprising at least one of the two available modes of prediction is constructed. For at least one mode of prediction included in the said list, termed the current mode, there is determined a measurement of distance between the predictor block associated with the said current mode and the other predictor blocks associated with the other available modes of prediction and obtained for the said current block. The list is then modified as a function of the said distance measurement determined. An information item identifying for the said current block, a mode of prediction from among the modified list is thereafter decoded on the basis of the coded data stream and the said current block and reconstructed on the basis of the predictor block associated with the said identified mode of prediction.