Measuring 3D face geometry for integration with apperance models
Siddharth K. Madan · Rutgers University Community Repository (Rutgers University) · 2008
Capturing the appearance of skin is a nontrivial problem. The appearance of skin depends significantly on the illumination and viewing direction. The global and fine scale local geometry influence the appearance of skin. In this thesis, we work towards capturing the appearance of skin by acquiring the global geometry and modeling the skin texture.We explore the use of structured light techniques to acquire global geometry. Twonovel structured light techniques have been developed. The techniques use image ratiosto obtain reliable reconstruction of low albedo regions. The use of laser scanner forgeometry acquisition and its advantages and disadvantages have also been discussed.We model the interaction of light with the skin surface as a combination of surface and subsurface scattering. The reflectance due to subsurface scattering and the reflectancedue to surface scattering have different properties and are modeled separately.The subsurface components lie on a lower dimensional linear subspace and the surfacecomponents lie on nonlinear manifolds. We look into the use of linear dimensionalityreduction techniques to model the subsurface reflectance and nonlinear dimensionalityreduction techniques to model the surface reflectance. In the initial results, separatemodeling of the surface and the subsurface components preserved the specularities.However, the specularities were lost when the reflectance were modeled without separating the surface and the subsurface components.