Local Illumination Compensation Flag Derivation in Video Coding

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing video coding technologies face challenges in efficiently deriving and storing the Local Illumination Compensation (LIC) flag for temporal candidates, leading to increased memory costs and computational complexity, especially when combined with merge mode encoding.

Innovation Solution

The techniques allow for the derivation of the LIC flag for temporal candidates from spatial neighbors, eliminating the need to store LIC flag information in a reference buffer by comparing POC values and matching reference pictures between spatial and temporal candidates.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If LIC flag information is stored in a reference buffer for temporal candidates, then LIC functionality can be properly implemented, but memory costs increase

Engineering Contradiction:
ImproveLIC functionalityVSAvoidmemory cost
Core Design Contradiction:
ReliabilityVSQuantity of substance

Solution Approach 1:

The patent extracts the LIC flag information from the reference buffer by deriving it from spatial neighbor blocks. Instead of storing LIC flags for temporal candidates in memory, the system retrieves spatial neighbors and derives the LIC flag information from them, eliminating the need for dedicated storage of this data in the reference buffer.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent copies LIC flag information from spatial neighbor blocks to be used for temporal candidates. Rather than storing and retrieving LIC flags from memory for each temporal candidate, the system copies the necessary information from already-available spatial neighbors, reducing memory access and storage requirements.

Inventive Principle:
Principle #26Copying

2Measurement precision

If LIC flag is derived for each temporal candidate, then LIC accuracy is improved, but computational complexity increases

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

Solution Approach 1:

The patent performs preliminary action by deriving LIC flags from spatial neighbors before they are needed for temporal candidate processing. The spatial neighbors are already decoded and available, so their LIC information is prepared in advance, eliminating the need for complex real-time derivation during temporal candidate processing.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The spatial neighbor blocks serve themselves by providing their LIC flag information to temporal candidates. Instead of requiring a separate complex derivation process for each temporal candidate, the system uses the self-contained LIC information from spatial neighbors, simplifying the overall computational process.

Inventive Principle:
Principle #25Self-service

Data Source

PatentEP3738310B1Video coding using local illumination compensation
Publication Date: 2025.01.08 QUALCOMM INC
  • EP3738310B1 patent drawingFigure 1
  • EP3738310B1 patent drawingFigure 2(a)~2(b)
  • EP3738310B1 patent drawingFigure 3(a)~3(b)

AI summary

A video coding device includes a memory configured to store video data, and processing circuitry in communication with the memory. The processing circuitry is configured to determine that a current block of the video data stored to the memory is to be coded according to a merge mode, and to determine a temporal candidate associated with a merge candidate list for the current block in accordance with the merge mode. The processing circuitry is further configured to determine a pool of spatial candidates associated with the merge candidate list, based on each spatial candidate of the pool having a same reference picture as the temporal candidate, and to set a local illumination compensation (LIC) flag value for the temporal candidate based on respective LIC flag values of one or more spatial candidates of the pool.