Image Decoding with Parallel Merge Regions for Motion Prediction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing video coding technologies face challenges in reducing complexity and improving encoding/decoding efficiency, particularly in high-resolution and high-quality image data transmission and storage.

Innovation Solution

A decoding apparatus and method that utilizes a parallel merge level to derive spatial merging candidates for current blocks, generating merging candidate lists based on merge flag information, and deriving motion information to improve encoding/decoding efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If conventional video coding techniques are used for high-resolution and high-quality image data, then image quality is improved, but transmission cost and storage cost are increased

Engineering Contradiction:
Improveimage qualityVSAvoidbit amount
Core Design Contradiction:
Manufacturing precisionVSQuantity of substance

Solution Approach 1:

The current block is divided into multiple parallel merging unit regions based on the parallel merge level. Each region independently derives spatial merging candidates, allowing parallel processing and reducing the overall complexity of the merging candidate generation process while maintaining high prediction accuracy for high-resolution images

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent derives spatial merging candidates from only certain neighboring blocks (lower left corner, left, upper right corner, upper, and upper left corner neighboring blocks) rather than all possible neighboring blocks. This partial action approach reduces the number of candidates to be processed while still providing sufficient prediction accuracy, thereby reducing bit amount without significantly compromising image quality

Inventive Principle:
Principle #16Partial or excessive action

2Measurement precision

If conventional merging candidate derivation methods are used, then prediction accuracy is improved, but computational complexity is increased

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

Solution Approach 1:

The LCU is divided into multiple parallel merging unit regions (MERs) based on the parallel merge level parameter. Each MER independently constructs its merging candidate list from its own neighboring blocks, enabling parallel derivation of spatial merging candidates across different regions. This segmentation reduces the overall computational complexity by distributing the work across multiple independent regions while maintaining prediction accuracy within each region

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent pre-defines the set of neighboring blocks to be used for deriving spatial merging candidates (lower left corner, left, upper right corner, upper, and upper left corner neighboring blocks). This preliminary identification of candidate sources streamlines the derivation process by eliminating the need to evaluate all possible neighboring blocks, thereby reducing computational complexity while maintaining sufficient prediction accuracy

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP4459997B1Image decoding apparatus, image encoding apparatus, and apparatus for transmitting data for an image
Publication Date: 2026.01.28 LG ELECTRONICS INC
  • EP4459997B1 patent drawingFigure 1
  • EP4459997B1 patent drawingFigure 2
  • EP4459997B1 patent drawingFigure 3

AI summary

A decoding apparatus for image decoding comprises a memory and at least one processor connected to the memory. The at least one processor is configured to receive information on a parallel merge level which represents a size of a parallel merging unit region and merge index information; obtain merge flag information which represents whether a merge mode is applied to a current block; derive spatial merging candidates for the current block based on the merge flag information indicating that the merge mode is applied to the current block; construct a merging candidate list including the spatial merging candidates for the current block; derive motion information of the current block based on one of the spatial merging candidates indicated by the merge index information in the merging candidate list; derive prediction samples of the current block based on the derived motion information; and generate a reconstructed picture based on the prediction samples. The size of the parallel merging unit region is derived based on the information on the parallel merge level. The current block belongs to the parallel merging unit region. The current block is related to a prediction unit, PU, and the PU is one of PUs partitioned from a coding unit, CU. The PU has a size smaller than the size of the parallel merging unit region and is located within the parallel merging unit region. The spatial merging candidates are identical to spatial merging candidates of a 2Nx2N PU which has a same size as the parallel merging unit region. The spatial merging candidates of the 2Nx2N PU which has the same size as the parallel merging unit region, are derived from a lower left corner neighboring block, a left neighboring block, an upper right corner neighboring block, an upper neighboring block and an upper left corner neighboring block of the parallel merging unit region. The information on the parallel merge level is received through a picture parameter set.