Intra Prediction Mode Derivation Using Inter-Predicted Block Samples

Solution Overview

Problem

Existing video encoding and decoding technologies do not optimally represent the characteristics of current blocks, leading to suboptimal performance in tools like combined inter and intra prediction (CIIP) and transform selection.

Innovation Solution

Derive intra prediction modes using inter prediction, block vector-based prediction, and texture analysis methods, such as intra template matching, to improve the accuracy of intra prediction modes by combining prediction histograms of gradients and template samples, and selecting the best modes based on distortion costs.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional intra prediction modes are used, then the encoding process is simple, but the prediction accuracy and representation of block characteristics are insufficient

Engineering Contradiction:

Improveprediction accuracyVSAvoidencoding complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent performs preliminary prediction using inter prediction or block vector-based prediction to generate prediction samples before deriving intra prediction modes. This preliminary action provides a foundation for more accurate intra mode derivation by using the prediction samples to analyze texture characteristics and determine gradient histograms, thereby improving prediction accuracy while managing complexity through a structured multi-step process

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent introduces prediction samples as an intermediary between the block being encoded and the intra prediction mode selection. These prediction samples, obtained through inter prediction or block vector-based prediction, serve as a mediator to analyze texture characteristics and derive gradient histograms, which then guide the selection of optimal intra prediction modes. This intermediary approach enables more accurate block characteristic representation

Inventive Principle:

Principle #24Intermediary (Mediator)

2Reliability

If more accurate intra prediction modes are derived using multiple methods, then the performance of CIIP and transform selection improves, but the computational complexity increases

Engineering Contradiction:

Improvetool performanceVSAvoidprocessing complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The patent performs preliminary prediction using inter prediction or block vector-based prediction to generate prediction samples before deriving intra prediction modes. This preliminary action provides a foundation for more accurate intra mode derivation by using the prediction samples to analyze texture characteristics and determine gradient histograms, thereby improving prediction accuracy while managing complexity through a structured multi-step process

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent applies different processing approaches to different regions or characteristics of the block. By analyzing texture characteristics locally through gradient histograms derived from prediction samples, the method selects intra prediction modes that are specifically suited to the local block characteristics. This local quality approach improves reliability for specific block types while avoiding unnecessary complexity for blocks that don't require advanced processing

Inventive Principle:

Principle #3Local quality

Data Source

PatentEP4664879A1Method and apparatus for obtaining one or more intra prediction modes (IPMS)

Publication Date: 2025.12.17 BEIJING XIAOMI MOBILE SOFTWARE CO LTD

Figure 1
Figure 2
Figure 3

AI summary

A method and an apparatus of processing one or more blocks of a picture obtain one or more intra prediction modes, IPMs, using one or more prediction samples of the block obtained by a certain prediction.