Parametric Reference Image Transformation for Video Coding

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The HEVC video coding standard faces challenges in achieving precise motion prediction due to the use of reference images that are not similar in texture and movement to the current image, leading to poor Inter coding performance.

Innovation Solution

A method is introduced that determines parameters for a parametric function to transform reference images into approximations that are more similar to the current image, using subsets of reference images that are temporally closest to the current image, thereby improving the accuracy of motion prediction and compression performance.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional reference images are used for inter prediction, then the coding process is simple, but the prediction precision deteriorates when reference images are not similar to the current image

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

Solution Approach 1:

The patent applies preliminary action by transforming reference images before they are used for prediction. The transformation is performed in advance on the reference images from the first subset to create approximations that better match the current image characteristics, thereby improving prediction precision before the actual coding process begins.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent employs parameter changes by applying a parametric function with determined parameters to transform the reference images. This function modifies the parameters of the reference images to make them more similar to the current image in terms of texture and movement characteristics, resolving the contradiction between simplicity and precision.

Inventive Principle:
Principle #35Parameter changes

2Productivity

If reference images with different texture and movement characteristics are used, then the coding process is faster, but the Inter coding performance deteriorates

Engineering Contradiction:
Improvecoding speedVSAvoidInter coding performance
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The patent performs preliminary transformation of reference images before they are used in the coding process. By pre-processing the reference images to match current image characteristics, the system maintains faster coding speed while improving the reliability of Inter coding performance through better-matched reference images.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent introduces an intermediary transformation process that acts as a mediator between the original reference images and the current image. This intermediary step creates approximated reference images that bridge the gap between different texture and movement characteristics, maintaining both coding efficiency and performance reliability.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Quantity of substance

If the current image is coded from temporally distant reference images, then the memory requirements are reduced, but the prediction accuracy deteriorates

Engineering Contradiction:
Improvememory requirementsVSAvoidprediction accuracy
Core Design Contradiction:
Quantity of substanceVSMeasurement precision

Solution Approach 1:

The patent applies parameter changes through a parametric transformation function that adapts temporally distant reference images to better match the current image. This transformation modifies the reference image parameters to compensate for temporal distance, maintaining prediction accuracy while allowing the use of fewer reference images in memory.

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent performs preliminary transformation on selected reference images to enhance their similarity to the current image before coding. This pre-processing step ensures that even temporally distant or fewer reference images can provide accurate predictions, reducing memory requirements while maintaining prediction accuracy.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP2901698B1Method of coding and decoding images, coding and decoding device and computer programs corresponding thereto
Publication Date: 2020.12.30 ORANGE SA
  • EP2901698B1 patent drawingFigure 1
  • EP2901698B1 patent drawingFigure 2~3A
  • EP2901698B1 patent drawingFigure 3B~5

AI summary

The invention relates to a method of coding at least one current image (In), characterized in that it comprises steps of: - determination (C4) of at least one parameter (p') of a predetermining function (Fp), said function being able to transform a first subset (SS) of a set (Sn) of previously decoded reference images into an approximation of a second subset (SC) of images of said set (Sn) of reference images, - application (C6) of said function (Fp) according to the parameter (p') determined to a third subset (SD) of said set (Sn) of reference images, said third subset being different from said first subset, so as to obtain another set (SV) of previously decoded reference images, - coding (C7) of the current image (In) on the basis of said obtained set (SV) of reference images.