| Author: | Yang, Di |
| Title: | Advancements in human skin color characterization : mechanisms, measurement, modeling, and applications |
| Advisors: | Wei, Minchen (BEEE) |
| Degree: | Ph.D. |
| Year: | 2026 |
| Department: | Department of Building Environment and Energy Engineering |
| Pages: | xviii, 160 pages : color illustrations |
| Language: | English |
| Abstract: | Human skin color plays a vital role in a wide range of applications, including digital imaging, display calibration, cosmetic product development, and dermatological assessment. However, accurately characterizing and reproducing skin color across diverse illumination and viewing conditions remains a complex challenge due to the interaction of physical properties, visual mechanisms, measurement constraints, and computational modeling limitations. This thesis addresses these challenges through four interrelated studies, each focusing on a different aspect: algorithm development, chromatic adaptation, measurement and modeling analysis, and the integration of experiments and simulations. The first study developed a color constancy algorithm based on the cone contrast color space, using datasets of real skin reflectance spectra and real light sources to construct reference gamut lookup tables. A loss function was designed to estimate illuminant chromaticities based on multiple gamut features. Experimental results demonstrated that the proposed algorithm consistently outperformed several existing algorithms, especially when image pixels were limited but dominated by skin colors. Furthermore, this study evaluated the influence of camera spectral sensitivity (CSS), revealing that measurement inaccuracies in CSS led to larger errors than those due to variation between similar devices. The second study examined how the interaction between adapting and stimulus illuminance affects the perceived display white point. Two experiments using full-screen image stimuli were conducted across 44 conditions, including a wide range of illuminance (30 to 15000 lx), CCTs (2500 to 10000 K), a dark condition, and an overcast condition. A lookup table (LUT) mapping adapting illumination to perceived white point CCT was generated. Validation showed that the LUT improved perceived attributes compared to default settings on commercial smartphones. Moreover, images with skin colors showed stronger correlations among comfort, naturalness, and preference, emphasizing the importance of accurate skin color reproduction. The third study analyzed two main techniques for measuring skin spectral reflectance. While spectrophotometers offered flexibility, they were affected by aperture size and measurement pressure, whereas spectroradiometers were sensitive to distance. Both methods were limited to fully separate specular and diffuse reflectance, making it difficult to capture accurate facial skin data. This study also reviewed skin modeling approaches, highlighting trade-offs between physical accuracy and computational efficiency. While simplified models reduce complexity by approximating skin layers and chromophores, they may sacrifice accuracy. In contrast, photon-based simulations offer higher fidelity but are computationally intensive. The fourth study integrated skin scales with skin models to support accurate skin color characterization. Representative skin scales were examined for target population, measurement methods, and color distribution in the CIELAB space. A skin model was constructed using both the Monte Carlo (MC) and Kubelka–Munk (KM) methods. The KM method achieved better performance in matching skin scale values and was used to build a LUT connecting skin color data to reflectance spectra. A psychophysical experiment further investigated the just-noticeable color difference (JNCD) using both patch and face stimuli under D65 and D95 white points. Results indicated that D95 led to less stable chromatic adaptation, with greater variability in ellipse orientation and size. Lighter skin colors consistently showed smaller JNCD ellipses, and patch stimuli yielded more consistent responses than face stimuli. In conclusion, this thesis presents an investigation into the mechanisms, measurement, modeling, and perceptual evaluation of human skin color. The four studies address challenges across theoretical and applied domains, offering insights and practical tools for enhancing color perception in imaging, display, and skin-related visual technologies. The integration of empirical data, psychophysical experiments, and physically based simulations supports future advances in color science, particularly in contexts where skin color appearance is important. |
| Rights: | All rights reserved |
| Access: | open access |
Copyright Undertaking
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item:
https://theses.lib.polyu.edu.hk/handle/200/14486