Image Encoding Sign Prediction Residual Coefficients

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current video coding methods incur high bit rate costs due to the systematic addition of sign bits in the data stream, which compromises signal compression performance and image reconstruction quality.

Innovation Solution

A method that predicts the sign of residual data coefficients based on a selected prediction mode, using a criterion that accounts for the amplitude and sign of residual data, to identify the most predictable signs and encode the differences, thereby optimizing compression performance.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If sign bits are systematically added to the data stream for encoding residual coefficients, then complete sign information is transmitted to the decoder, but bandwidth consumption increases and compression performance deteriorates

Engineering Contradiction:
Improvesign information completenessVSAvoidbit rate
Core Design Contradiction:
ReliabilityVSQuantity of substance

Solution Approach 1:

The patent extracts only the necessary sign information from the residual coefficients. Instead of transmitting all sign bits systematically, it identifies and transmits only the signs of significant coefficients (those above a threshold), removing redundant sign information from the data stream while maintaining decoding accuracy.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent changes the parameter being transmitted from complete sign information to selective sign information based on coefficient magnitude. By using a threshold parameter to determine which signs to transmit, it transforms the encoding approach from universal sign transmission to conditional sign transmission, reducing bit rate while preserving essential information.

Inventive Principle:
Principle #35Parameter changes

2Measurement precision

If all residual coefficient signs are encoded explicitly, then decoding accuracy is maintained, but compression efficiency decreases

Engineering Contradiction:
Improvedecoding accuracyVSAvoidcompression efficiency
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent applies different encoding treatments to different coefficients based on their local properties (magnitude). Significant coefficients (above threshold) receive explicit sign encoding to maintain precision, while insignificant coefficients (below threshold) have their signs omitted, as they contribute minimally to reconstruction accuracy. This local differentiation optimizes the balance between precision and efficiency.

Inventive Principle:
Principle #3Local quality

3Quantity of substance

If sign prediction is applied to all coefficients, then bandwidth is reduced, but disturbances are introduced that impair compression efficiency

Engineering Contradiction:
Improvedata stream sizeVSAvoidcompression efficiency disturbance
Core Design Contradiction:
Quantity of substanceVSObject-generated harmful factors

Solution Approach 1:

Instead of applying sign prediction to all coefficients (excessive action), the patent applies it only to significant coefficients that meet a magnitude threshold (partial action). This selective application avoids introducing prediction disturbances in coefficients where such disturbances would have minimal impact, thereby maintaining compression efficiency while still achieving bandwidth reduction.

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP3061246B1Method for encoding and decoding images, device for encoding and decoding images and corresponding computer programs
Publication Date: 2022.04.13 ORANGE SA
  • EP3061246B1 patent drawingFigure 1
  • EP3061246B1 patent drawingFigure 2
  • EP3061246B1 patent drawingFigure 3~4

AI summary

The invention relates to a method for encoding at least one image (ICj) cut into blocks, implementing, for a current block to be encoded, the steps of: predicting (C3) the current block in accordance with a prediction mode selected among a plurality of predetermined prediction modes (MP0, MP1,…, MPv,…, MPQ)M; obtaining (C3) a predictor block (BPopt); determining (C4-C7) a residual data set that is representative of the difference between the predictor block obtained and the current block, said residual data being likely to have an amplitude and a sign. According to the invention, the encoding method performs the following steps: selecting (C9), from said residual data set, at least one piece of residual data assigned an amplitude and a sign, by means of a predetermined criterion which is a function of the selected prediction mode; obtaining (C11) a predicted value of the sign of said at least one selected piece of residual data; calculating (C12) information (Id) representing the difference between the predicted value of said sign and the value of the sign of said at least one selected piece of residual data; encoding (C13) said calculated information.