Multi-illuminant Color Correction via Segmented Illuminant Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing methods struggle with color correction in scenes illuminated by multiple sources of light, as they assume uniform illumination from a single source, leading to poor performance when faced with complex lighting conditions.
Innovation Solution
A machine learning model is trained to detect multiple sources of light and generate estimated illuminant images, which include the color and contribution of each source, allowing for the combination of light sources to correct the image, even when the actual number of sources differs from the assumed number.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a machine learning model is trained to detect multiple sources of light and generate separate estimated illuminant images for each source, then color correction performance in multi-illuminant scenes is improved, but device complexity and computational resources increase
Solution Approach 1:
The patent segments the illumination estimation task by generating separate estimated illuminant images for each detected light source rather than producing a single combined illumination estimate. This segmentation allows the system to handle multiple illuminants independently, improving color correction accuracy in complex lighting scenarios while managing computational complexity through structured processing of individual source contributions
Solution Approach 2:
The patent introduces an additional dimension to the illumination estimation by adding the temporal dimension through video sequences. By leveraging temporal information from multiple frames, the system can better distinguish between multiple light sources and improve illumination estimation accuracy without proportionally increasing spatial computational requirements in individual frames
2Measurement precision
If the machine learning model generates high-resolution estimated contribution images for each light source, then the precision of illumination estimation is improved, but the computational load and processing time increase significantly
Solution Approach 1:
The patent applies partial action by generating estimated contribution images at reduced resolution rather than full image resolution. This approach provides sufficient precision for illumination estimation and color correction purposes while significantly reducing computational load and processing time. The reduced resolution is adequate because the contribution images serve to weight and combine illuminant estimates rather than require pixel-perfect accuracy
3Ease of manufacture
If the machine learning model is trained to detect a fixed number N of light sources, then the model structure and training process are simplified, but the model may not accurately handle scenes with a different actual number of light sources
Solution Approach 1:
The patent implements universality by designing the machine learning model to detect a fixed number N of light sources that can accommodate varying actual numbers of illuminants in different scenes. The model structure remains consistent with the same number of output channels, but it can adapt to handle cases where some light sources may be dominant in certain regions or frames, providing versatile performance across diverse multi-illuminant scenarios without requiring multiple specialized models
Data Source
AI summary
The present disclosure relates to an image processing method for correcting colors in an input image representing a scene, the image processing method including: processing the input image with a machine learning model, wherein the machine learning model is previously trained to detect a predefined number N>1 of sources of light illuminating the scene and to generate N estimated illuminant images associated respectively to the N sources of light, wherein each estimated illuminant image includes an estimated color of the light emitted by the respective source of light and an estimated contribution image; generating a total illuminant image by using the N estimated illuminant images; and generating an output image by correcting the colors in the input image based on the total illuminant image.