3D object shape and reflectance property reconstruction using image scanner

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 constructed by two processes. First is the initial shape estimation and second is the iterative process to estimate the object shape, color and specular components. In this paper, we propose two methods for the initial shape estimation: one is the use of multiple light sources along circular paths and the other is the use of two groups of two light sources located in same direction. From the experimental results using synthetic images, multiple light sources can be estimate object shape, color and specular reflections correctly. But, because of many light sources, the structure of the image scanner will be complex. On the other hand, in case of two groups of two light sources, estimated object color and specular components has some errors but the object shape can be estimated accurately.