Shape reconstruction using extended photometric stereo

Abstract

The main interest of previous researches for shape from shading was only on Lambertian surfaces which consists of diffuse reflectance surfaces. In practice, natural objects have hybrid reflectance, which limits the recovery of shape from intensity distribution. We propose a method of determining the shape using a neural network and diffuse illumination. The segmented region of the sphere surface is used for the training set, which can be determined by the shadow line and the surface edge. Diffuse illumination is used to avoid specular spikes and highlights which usually cause many problems such as intensity disparities. The diffuse illumination method using flat paper allows us to reduce these phenomena with a simple scheme. The neural network and diffuse illumination method is useful for shape from shading, because it can be applied to objects of unknown reflectance properties, but it is applied only to Torrance-Sparrow (1967) style reflectance.