Extrapolation-Based Intra Prediction Using Adaptive Filter Coefficients

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing video coding techniques face challenges in efficiently utilizing reconstructed reference samples and predicted signals for intra prediction, leading to suboptimal video coding efficiency and quality, particularly with increasing image sizes and resolutions.

Innovation Solution

A method and apparatus that utilize an extrapolation filter to derive filter coefficients based on a rectangular reference sample region and shape for intra prediction, applying these coefficients to both reconstructed reference samples and predicted samples to generate prediction blocks.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If traditional intra prediction methods are used, then the coding process is simple, but video coding efficiency and quality are insufficient for high-resolution images

Engineering Contradiction:
Improvevideo coding efficiencyVSAvoidprediction process complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent segments the prediction process by separating reference sample selection, filter coefficient derivation, and extrapolation filtering into distinct stages. The current block is divided into multiple prediction blocks that can be processed independently, allowing parallel computation and improving coding efficiency without overwhelming system complexity.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent performs preliminary actions by pre-deriving filter coefficients from reference samples before the actual prediction process. The reference sample regions are identified and prepared in advance, and filter coefficients are calculated beforehand, which streamlines the main prediction process and improves overall coding efficiency.

Inventive Principle:
Principle #10Preliminary action

2Measurement precision

If more reference samples are used for prediction, then prediction accuracy improves, but computational complexity and processing time increase

Engineering Contradiction:
Improveprediction accuracyVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent applies local quality by using different reference sample regions for different prediction blocks based on their specific characteristics. Each prediction block uses a locally optimized set of reference samples and filter coefficients tailored to its position and content, improving prediction accuracy without requiring all reference samples to be processed uniformly.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent uses partial action by selectively applying extrapolation filtering to only those prediction blocks where it provides significant improvement, rather than uniformly applying it to all blocks. This selective approach maintains high prediction accuracy for complex regions while reducing processing time for simpler regions.

Inventive Principle:
Principle #16Partial or excessive action

3Manufacturing precision

If extrapolation filtering is applied to all prediction blocks, then prediction quality improves, but computational load and hardware resource requirements increase

Engineering Contradiction:
Improveprediction qualityVSAvoidhardware resource requirements
Core Design Contradiction:
Manufacturing precisionVSDevice complexity

Solution Approach 1:

The patent introduces dynamics by making the application of extrapolation filtering adaptive rather than static. The filtering process is dynamically applied based on block characteristics, reference sample availability, and predicted quality improvement, allowing the system to optimize between quality and computational load in real-time during encoding.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The patent changes parameters by adjusting filter coefficients, reference sample region sizes, and extrapolation filter application thresholds based on local image characteristics. These parameter changes allow the system to maintain high prediction quality where needed while reducing computational complexity in regions where simpler prediction suffices.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP4697721A1Method for extrapolation-based intra prediction
Publication Date: 2026.02.18 HYUNDAI MOTOR CO LTD
  • EP4697721A1 patent drawingFigure 1
  • EP4697721A1 patent drawingFigure 2
  • EP4697721A1 patent drawingFigure 3A

AI summary

A method and an apparatus are disclosed for video coding using an extrapolation-based intra prediction. In the disclosed embodiments, the video decoding apparatus determines a reference sample region used for deriving filter coefficients. Here, the reference sample region includes reconstructed reference samples of the current block. The video decoding apparatus determines a filter shape used for an extrapolation-based intra prediction. Here, the filter shape is rectangular and includes a prediction pixel region and an input pixel region. The video decoding apparatus derives the filter coefficients based on the reference sample region and the filter shape. The video decoding apparatus generates a prediction block of the current block by applying the derived filter coefficients to the reconstructed reference samples of the current block and predicted samples.