Image Prediction Using Reference Pixels and Affine Sub-Blocks

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing image compression technologies face challenges in efficiently performing intra prediction based on reference pixel availability, inter prediction in units of sub-blocks, and determining affine candidates.

Innovation Solution

The method involves checking a reference pixel region, determining a reference pixel processing setting based on availability, generating a candidate list for motion information prediction, deriving a control point vector, and performing intra and inter prediction using motion vectors, with affine candidates including spatial, temporal, and constructed candidates, and motion vectors derived in units of sub-blocks.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If intra prediction is performed based on reference pixel availability, then coding performance is improved, but device complexity increases due to additional availability checking and processing setting determination

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

Solution Approach 1:

The patent checks reference pixel availability and determines processing settings in advance before performing intra prediction. This preliminary action allows the system to prepare the optimal prediction approach based on available reference data, improving coding efficiency without adding complex runtime decisions.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent dynamically adjusts the intra prediction processing based on the availability of reference pixels. The system adapts its prediction strategy according to the determined reference pixel processing setting, allowing flexible optimization of coding performance for different block types and reference availability scenarios.

Inventive Principle:
Principle #15Dynamics

2Measurement precision

If inter prediction is performed in units of sub-blocks with affine candidates, then inter prediction accuracy is improved, but encoding/decoding time increases due to additional candidate generation and motion vector derivation

Engineering Contradiction:
Improveinter prediction accuracyVSAvoidencoding/decoding time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent divides the current block into sub-blocks and performs inter prediction independently for each sub-block using affine motion models. This segmentation allows more precise motion compensation for different regions while maintaining manageable computational complexity through localized processing.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent changes the motion modeling parameters by introducing affine transformation parameters (control point vectors, motion vectors) that describe motion more accurately than traditional block-based methods. This parameter change enables higher prediction accuracy by modeling rotational, scaling, and shearing motions.

Inventive Principle:
Principle #35Parameter changes

3Measurement precision

If multiple affine candidates including spatial, temporal, and constructed candidates are generated, then motion information prediction accuracy is improved, but device complexity increases due to expanded candidate list management

Engineering Contradiction:
Improvemotion information prediction accuracyVSAvoidcandidate list complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent merges multiple types of motion candidates (spatial, temporal, and constructed candidates) into a unified candidate list. This combination allows the system to leverage diverse motion information sources, improving prediction accuracy by selecting from a broader range of motion models.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The patent introduces constructed candidates as intermediary motion models that combine information from spatial and temporal neighbors. These constructed candidates act as mediators that synthesize motion information from multiple sources, providing additional accurate prediction options without requiring separate processing paths.

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentUS12519933B2Image encoding/decoding method and device
Publication Date: 2026.01.06 INST OF IMAGE TECH INC
  • US12519933B2 patent drawing
  • US12519933B2 patent drawing
  • US12519933B2 patent drawing

AI summary

An image prediction method according to the present invention may comprise: identifying a reference pixel region designated for obtaining correlation information; determining a reference pixel processing configuration on the basis of determination of the availability of the reference pixel region; and performing intra prediction according to the determined reference pixel processing. As described above, performing intra prediction on the basis of the availability of a reference pixel according to the present invention can improve encoding performance.