Multi-Reference-Line Intra Prediction With Distance-Weighted Samples

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The increasing demand for high-resolution and high-quality images leads to a rise in transmission and storage costs due to the increased amount of information, necessitating a high-efficiency image compression technology.

Innovation Solution

An image encoding/decoding method and apparatus utilizing Multi Reference Lines (MRL) for intra prediction, which generates prediction blocks through a weighted sum of reference samples and fuses multiple prediction blocks, including planar modes within the Template based Intra Mode Derivation (TIMD) mode.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If high-resolution and high-quality image data is transmitted and stored, then image quality is improved, but transmission and storage costs increase

Engineering Contradiction:
Improveimage qualityVSAvoidamount of information
Core Design Contradiction:
Manufacturing precisionVSQuantity of substance

Solution Approach 1:

The patent applies parameter changes by using multiple reference lines instead of a single reference line for intra-prediction, and by employing weighted sum operations with distance-based weights. This changes the prediction parameters to achieve better compression efficiency for high-resolution images, reducing the amount of information that needs to be transmitted and stored while maintaining image quality

Inventive Principle:
Principle #35Parameter changes

2Measurement precision

If multiple reference lines are used for intra prediction, then prediction accuracy is improved, but computational complexity increases

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

Solution Approach 1:

The patent applies local quality by using different weights for different reference samples based on their distance from the current block. Closer reference samples receive higher weights while farther samples receive lower weights. This localized weighting approach improves prediction accuracy for specific regions without uniformly increasing complexity across the entire image

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent changes the prediction parameters by incorporating distance-based weighting in the weighted sum operation. Instead of treating all reference samples equally, the method uses distance parameters to dynamically adjust weights, improving prediction accuracy while keeping the computational approach systematic and manageable

Inventive Principle:
Principle #35Parameter changes

3Measurement precision

If weighted sum of reference samples is used, then prediction precision is improved, but encoding complexity increases

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

Solution Approach 1:

The patent applies parameter changes by using distance-based weights in the weighted sum operation. The weight for each reference sample is determined by its distance parameter, creating a systematic approach to weighting that improves prediction precision while maintaining encoding efficiency through parameterized weight calculation

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP4593377A1Intra prediction mode-based image encoding/decoding method and apparatus using multi reference line (MRL), and recording medium for storing bitstream
Publication Date: 2025.07.30 LG ELECTRONICS INC
  • EP4593377A1 patent drawingFigure 1
  • EP4593377A1 patent drawingFigure 2
  • EP4593377A1 patent drawingFigure 3

AI summary

An image encoding/decoding method and device are provided. An image decoding method according to the present disclosure comprises the steps of: determining reference samples separated by a distance of n samples from a current block and reference samples separated by a distance of m samples from the current block (wherein the reference samples separated by the distance of n samples include a top reference sample and a top-right reference sample, and reference samples separated by the distance of m samples include a left reference sample and a bottom-left reference sample); and generating a prediction block of the current block on the basis of a weighted sum of at least two of the determined reference samples, wherein the weight used in the weighted sum calculation is determined on the basis of the n or m, which may be natural numbers.