3D shape scanner using multiple light sources

Abstract

This study proposes a method to recover 3D shape, color and specular reflections of an object from images taken by an image scanner which has multiple illuminations. The proposed method is based on the photometric stereo method and it is constructed by several processes. First is the initial shape estimation using four images, and then the color and specular component estimation are performed sequentially. Finally, we apply the non-linear least square method to estimate optimal parameters of shape, color and specular components. In experiments, we perform the object reconstruction using synthetic images. Estimated color and specular components of object has some errors but the object shape can be estimated accurately.