Context-Aware Palette Extraction for Adaptive Lighting Control

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing lighting control systems fail to adapt the image-based control of light sources to the specific environment they are installed in, leading to unsuitable lighting conditions.

Innovation Solution

A method and system that obtain contextual information about the environment to modify a source image, extract colors from the modified image, and control light sources based on these colors, ensuring the lighting adapts to the environment.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Speed

If k-means clustering is used for palette extraction, then computational speed is improved, but color accuracy deteriorates due to inability to handle contextual color relationships

Engineering Contradiction:
Improvecomputational speedVSAvoidcolor accuracy
Core Design Contradiction:
SpeedVSMeasurement precision

Solution Approach 1:

The patent segments the image processing into multiple stages: initial palette generation using k-means clustering, followed by contextual refinement using graph convolutional networks. This segmentation allows the system to benefit from the speed of k-means while adding accuracy through subsequent contextual analysis of color relationships in the image.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces an intermediary contextual analysis layer between the initial k-means clustering and the final palette output. This intermediary layer uses graph convolutional networks to analyze color relationships and refine the palette, acting as a mediator that improves accuracy without completely replacing the efficient k-means approach.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Use of energy by moving object

If traditional color quantization methods are used, then computational resources are conserved, but handling of large high-resolution images becomes inefficient

Engineering Contradiction:
Improvecomputational resourcesVSAvoidprocessing efficiency
Core Design Contradiction:
Use of energy by moving objectVSProductivity

Solution Approach 1:

The patent divides the image processing task into segmentation steps where the image is first divided into superpixels or patches, then processed in smaller manageable units. This segmentation allows efficient handling of large high-resolution images by processing local regions independently before combining results, reducing overall computational resource requirements while maintaining productivity.

Inventive Principle:
Principle #1Segmentation

3Device complexity

If context is ignored in palette extraction, then processing complexity is reduced, but semantic color relationships are lost

Engineering Contradiction:
Improveprocessing complexityVSAvoidsemantic color relationships
Core Design Contradiction:
Device complexityVSLoss of information

Solution Approach 1:

The patent performs preliminary contextual analysis by building a graph representation of color relationships before final palette determination. This preliminary action captures semantic color relationships early in the process, allowing the main palette extraction to proceed with this contextual information already integrated, rather than requiring complex real-time contextual processing.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP4466947B1Modification of images for palette extraction based on context
Publication Date: 2026.05.06 SIGNIFY HOLDING BV
  • EP4466947B1 patent drawingFigure 1
  • EP4466947B1 patent drawingFigure 2
  • EP4466947B1 patent drawingFigure 3

AI summary

A method (300) of controlling a plurality of light sources (140, 142) based on a source image (200) is provided. The plurality of light sources (140, 142) are installed in an environment (150). The method (300) comprises: obtaining (302) contextual information of the environment (150), modifying (304) the source image (200) based on the contextual information into a modified source image (210), extracting (306) a set of colors present in the modified source image (210), and controlling (314) the plurality of light sources (140, 142) based on the set of colors of the modified source image (210). A lighting control system (100) and a computer program product for such a system (100) are also provided.