Intra Image Coding Using Residual Prediction to Cut Signaling Bits

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing image coding methods require efficient reduction of bits for residual signals and intra prediction modes, particularly in intra prediction, to enhance coding efficiency.

Innovation Solution

Applying residual prediction to intra prediction by determining reference blocks and signals based on neighboring blocks in the current picture, reducing the need for signaling intra prediction modes and optimizing bit usage.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of substance

If conventional intra prediction coding is used, then the coding process is simple, but the number of bits required for residual signals is large

Engineering Contradiction:
Improvenumber of bits for residual signalsVSAvoidcoding process complexity
Core Design Contradiction:
Loss of substanceVSDevice complexity

Solution Approach 1:

The patent applies preliminary action by performing residual prediction using reference blocks before final residual coding. The encoder and decoder both pre-compute predicted residual signals from neighboring blocks, allowing the actual residual to be coded with fewer bits since it represents only the difference from the predicted residual rather than the full residual signal.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent introduces an intermediary element - the predicted residual signal derived from reference blocks - that mediates between the original residual signal and the final coded representation. This intermediary allows the residual to be split into a predictable component (coded with fewer bits) and a smaller difference component.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Loss of substance

If intra prediction modes are signaled explicitly, then the decoder can accurately reconstruct the prediction, but the number of bits required for signaling increases

Engineering Contradiction:
Improvenumber of bits for signaling intra prediction modeVSAvoidintra prediction mode accuracy
Core Design Contradiction:
Loss of substanceVSMeasurement precision

Solution Approach 1:

The patent applies self-service by enabling the decoder to determine the intra prediction mode independently using the same reference block and calculation method as the encoder. Both sides perform identical computations on the same reference data, allowing the prediction mode to be implicitly agreed upon without requiring explicit signaling bits.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

Instead of the conventional approach where the encoder selects and signals the prediction mode to the decoder, the patent inverts the process by having both encoder and decoder independently derive the prediction mode from reference blocks. This reversal eliminates the need for mode signaling while maintaining prediction accuracy.

Inventive Principle:
Principle #13The other way round (Inversion)

3Productivity

If more reference blocks are used for residual prediction, then the coding efficiency improves, but the computational complexity increases

Engineering Contradiction:
Improvecoding efficiencyVSAvoidcomputational complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent applies local quality by using multiple reference blocks with different characteristics (e.g., different spatial locations, different prediction directions) to predict different portions or aspects of the residual signal. Each reference block is selected and used based on its local suitability for predicting specific residual patterns, optimizing the trade-off between prediction accuracy and computational effort.

Inventive Principle:
Principle #3Local quality

Data Source

PatentEP4723622A1Image coding method and device therefor
Publication Date: 2026.04.08 SAMSUNG ELECTRONICS CO LTD
  • EP4723622A1 patent drawingFigure 1
  • EP4723622A1 patent drawingFigure 2
  • EP4723622A1 patent drawingFigure 3

AI summary

The disclosure relates to an image decoding method, and more particularly, to a method comprising: obtaining a residual difference signal of the current block by processing a bitstream; determining a reference block corresponding to the current block, in the current picture including the current block; determining a residual signal of a reference block, based on a reconstructed signal of the reference block and a prediction signal of the reference block; and determining a reconstructed signal of the current block, based on a residual difference signal of the current block, a residual signal of the reference block, and a prediction signal of the current block, wherein the prediction signal of the reference block is determined based on at least one reference pixel included in a neighboring block of the reference block in the current picture, and the prediction signal of the current block is determined based on at least one reference pixel included in a neighboring block of the current block in the current picture. The disclosure also relates to an apparatus for performing the method.