Separating Subsurface Scattering from Photometric Image
Tai-Pang Wu, Chi–Keung Tang · 2006
While subsurface scattering is common in many real objects, almost all separation algorithms focus on extracting specular and diffuse components from real images. In this paper, we present a model-less approach derived from the bi-directional surface scattering reflectance distribution function (BSSRDF). In our approach, we show that an illumination image is composed by the Lambertian diffuse and subsurface scattering images. By converting the separation problem into one of two-layer separation in the illumination domain, a Bayesian framework is used to solve the optimization problem which incorporates spatial and illumination constraints, the latter of which are captured as a set of diffuse priors. We present the detailed mathematical formulation and experimental results.