Geometry-Partitioned Image Prediction With Neighbor Compensation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The increasing demand for high-resolution and high-quality images leads to higher transmission and storage costs due to increased bit rates, necessitating a high-efficiency image compression technology.

Innovation Solution

An image encoding/decoding method that applies a geometry partitioning mode with compensation, deriving compensation information based on partitioning direction and offset, and signaling whether compensation is applied, stored in a non-transitory computer-readable recording medium for image reconstruction.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If high-resolution and high-quality images are transmitted, then image quality is improved, but transmission cost and storage cost increase due to increased bit rate

Engineering Contradiction:
Improveimage qualityVSAvoidtransmission cost and storage cost
Core Design Contradiction:
Measurement precisionVSLoss of energy

Solution Approach 1:

The patent applies illumination compensation by deriving compensation information (scaling factor and offset) from neighboring blocks and applying it to reference blocks. This changes the parameters of the reference block to better match the current block's illumination conditions, improving prediction accuracy and reducing residual energy, thereby achieving better compression efficiency without sacrificing image quality

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent performs preliminary illumination compensation on reference blocks before using them for prediction. By pre-adjusting the reference blocks based on neighboring block characteristics, the system prepares optimized prediction data in advance, which reduces the amount of data that needs to be transmitted and stored while maintaining high image quality

Inventive Principle:
Principle #10Preliminary action

2Measurement precision

If geometry partitioning mode is applied to a current block, then prediction accuracy is improved, but complexity of deriving and signaling compensation information increases

Engineering Contradiction:
Improveprediction accuracyVSAvoidcomplexity of deriving and signaling compensation information
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent applies different compensation strategies to different regions defined by geometry partitioning. By dividing the current block into multiple regions with different motion characteristics and applying illumination compensation locally to each region, the system achieves higher prediction accuracy while managing complexity through region-specific processing

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent derives compensation information for the current block by copying and adapting information from neighboring blocks. This copying approach reduces the complexity of deriving compensation information from scratch, as the system leverages already-available neighboring block data to generate compensation parameters for the current block

Inventive Principle:
Principle #26Copying

3Measurement precision

If compensation information is derived from neighboring area of reference block and current block, then prediction accuracy is improved, but encoding complexity increases

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

Solution Approach 1:

The patent uses neighboring blocks to serve multiple functions: they provide both motion information and illumination compensation information. By deriving scaling factors and offsets from neighboring block characteristics, the system achieves multi-functional use of available data, improving prediction accuracy without proportionally increasing encoding complexity

Inventive Principle:
Principle #6Universality (Multi-functionality)

Data Source

PatentEP4683327A1Image encoding/decoding method, method for transmitting bitstream, and recording medium having bitstream stored therein
Publication Date: 2026.01.21 LG ELECTRONICS INC
  • EP4683327A1 patent drawingFigure 1
  • EP4683327A1 patent drawingFigure 2
  • EP4683327A1 patent drawingFigure 3

AI summary

Provided are an image encoding/decoding method, a bitstream transmission method, and a computer-readable recording medium storing a bitstream. The image decoding method performed by an image decoding apparatus, according to the present disclosure, comprises the steps of: acquiring at least one piece of motion information regarding the current block to which a geometry partitioning mode (GPM) is applied; on the basis of a neighboring region of at least one reference block indicated by the motion information and a neighboring region of the current block, deriving compensation information related to compensation of the reference block; and generating prediction blocks of the current block on the basis of the compensation information.