Video Block Intra Prediction Recalibration for Future Coding Efficiency

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current video compression standards face inefficiencies in intra prediction due to inaccurate intra prediction modes (IPMs) and signaling overhead, which affect the performance of future blocks relying on previously coded blocks, particularly in tools like MPM and OBIC.

Innovation Solution

A method to recalibrate intra prediction modes (RIPMs) by determining a list of recalculated RIPMs based on the reconstructed block, involving distortion cost calculation and sorting to select optimal modes, which can be used for future blocks, and incorporating this into encoding and decoding processes.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If regular intra prediction modes are used for coding future blocks, then encoding simplicity is maintained, but prediction accuracy deteriorates due to reliance on potentially inaccurate neighboring block IPMs

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

Solution Approach 1:

The patent applies preliminary action by recalculating intra prediction modes for the current block before they are used by future blocks. The encoder/decoder determines recalculated RIPMs based on the reconstructed block of the current block, so that when future blocks need intra prediction modes from neighboring blocks, they already have accurate recalculated modes available. This advance recalculation ensures prediction accuracy is improved before the modes are propagated to future blocks.

Inventive Principle:
Principle #10Preliminary action

2Measurement precision

If accurate intra prediction modes are selected for current block, then prediction quality improves, but signaling overhead increases

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

Solution Approach 1:

The patent implements feedback by using the reconstructed block of the current block to recalculate intra prediction modes. The distortion cost is calculated between the reconstructed block and prediction blocks generated by different RIPMs, and this feedback information is used to select the optimal RIPMs. This closed-loop approach ensures that the selected modes accurately represent the actual block content rather than relying on potentially inaccurate neighboring block modes.

Inventive Principle:
Principle #23Feedback

3Reliability

If recalculated RIPMs are determined for every current block, then prediction accuracy for future blocks improves, but processing time increases

Engineering Contradiction:
Improveintra prediction performanceVSAvoidprocessing time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The patent applies parameter changes by dynamically adjusting which RIPMs are included in the recalculation based on block characteristics. The first list of RIPMs can be configured to include all RIPMs or a subset depending on the specific application requirements. Additionally, the distortion cost calculation and mode selection parameters can be adjusted to balance between accuracy and processing time, allowing flexible adaptation to different performance requirements.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP4679822A1Encoding/decoding video picture data
Publication Date: 2026.01.14 BEIJING XIAOMI MOBILE SOFTWARE CO LTD
  • EP4679822A1 patent drawingFigure 1~2
  • EP4679822A1 patent drawingFigure 3~4
  • EP4679822A1 patent drawingFigure 5

AI summary

The present disclosure relate to prediction of blocks of a video. For a current block of video picture, coded in any mode, the present invention calculates a list of at least one RIPM based on reconstructed samples of said current block. The list of the at least one RIPM is then stored in memory and is used for coding future blocks in the video for example through MPM or OBIC modes. So, the calculated RIPMs for the current block are not changing the sample reconstruction of the current block but rather making the coding of future blocks to be more efficient.