Exploring Exploring Independent Component Analysis for Skin Tone Characterization
Session Number
1
Advisor(s)
Nicole Tin, Velexi Research
Location
A133
Discipline
Computer Science
Start Date
15-4-2026 10:15 AM
End Date
15-4-2026 11:00 AM
Abstract
A color space is an organization and representation of colors in terms of specified components. While RGB is the most common, it is not less useful when looking at how humans perceive skin tones under different lighting conditions. Other color spaces, like CIELAB and HSV, are developed to separate lightness and chromatic information. However, these color spaces are still based on predefined mathematical structures. Instead of choosing one of these predefined color spaces and analyzing skin tone data sets, ICA, an independent dimension reduction method, is applied to create a skin tone color space based on a predefined cosmetic shade scale (L’Oreal). This project explores, through Python-based analysis and 3D visualization, if ICA can create an even more consistent and understandable skin tone color space compared with traditional color spaces. The proposed research applies ICA, an independent dimension reduction method, instead of traditional color spaces, and explores if ICA can create an even more consistent and comprehensive skin tone color space.
Exploring Exploring Independent Component Analysis for Skin Tone Characterization
A133
A color space is an organization and representation of colors in terms of specified components. While RGB is the most common, it is not less useful when looking at how humans perceive skin tones under different lighting conditions. Other color spaces, like CIELAB and HSV, are developed to separate lightness and chromatic information. However, these color spaces are still based on predefined mathematical structures. Instead of choosing one of these predefined color spaces and analyzing skin tone data sets, ICA, an independent dimension reduction method, is applied to create a skin tone color space based on a predefined cosmetic shade scale (L’Oreal). This project explores, through Python-based analysis and 3D visualization, if ICA can create an even more consistent and understandable skin tone color space compared with traditional color spaces. The proposed research applies ICA, an independent dimension reduction method, instead of traditional color spaces, and explores if ICA can create an even more consistent and comprehensive skin tone color space.