Multi-spectral Imaging Method for
3D Computer Graphics Reproduction of Human Skin
This paper uses a multi-spectral imaging method to develop a measurement system for three-dimensional computer graphics (3D CG) reproduction of human skin under various scene illumination such as natural daylight. The reflection property of human skin is modeled based on a multi-spectral reflection model. A variable angle measuring device was developed for measuring the reflectance properties of the human skin. Variable angle images of a human face taken by an RGB camera were used to estimate spectral reflectance. Then, spatial and spectral distributions of various scene illumination were measured using an RGB digital camera with a fish-eye lens for illumination information for rendering by 3D CG. Finally, 3D CG images of human skin were reproduced with the estimated reflection model parameter and the spectral distribution of scene illumination. Our results demonstrate that multi-spectral information is necessary for more accurate 3D CG rendering of human skin.