Accurate 3D Face and Body Modeling from a Single Fixed Kinect

Ruizhe Wang*, Matthias Hernandez*, Jongmoo Choi, Gérard Medioni Computer Vision Lab, IRIS University of Southern California Abstract In this paper, we address the problem of both face and body modeling using a single fixed low-cost 3D camera (e.g. Kinect). Unlike other scanning technologies which either set up multiple sensors around a static subject or scan the static subject with a hand-held moving 3D camera, our method allows the subject to move in front of a single fixed 3D sensor. This opens the door to a wide range of applications where scanning is conveniently performed at home alone. While partial range scans of face are aligned rigidly, the body modeling is performed through articulated registration. At the surface reconstruction stage, we utilize the cylindrical representation for smoothing, hole filling and blending. Experimental results demonstrate the effectiveness of our modeling technique.