3-D Reconstruction of Multipart Self-Occluding Objects

In this paper we present a method for reconstruction of multipart objects from several arbitrary views using deformable superquadrics as the models of the object's parts. Two visual cues are used: occluding contours and stereo (possibly aided by projected patterns). The object can be relatively complex and can exhibit numerous self occlusions from some or all views. Our preliminary experiments on a human body and a tailor's mannequin show that the reconstruction is more complete than in purely stereo or structured light based methods and more precise than the reconstruction from occluding contours only.