Shape reconstruction from shadow and shading

This paper presents a method of reconstructing convex polyhedral objects by the "shape from shadow" approach. The process utilizes several gray scale images captured by a fixed camera with six light sources at different positions. The positions of the camera and light sources are calibrated. The two-dimensional shadows casted by an object are extracted from the images. The geometrical relationship between the light sources, the camera, and the corners of the shadow boundary are used to obtain three-dimensional coordinates of the corners of the polyhedral in a world coordinate frame. A volumetric representation of the shape of an object is then possible.<>