3-D Modeling of Human Heads from Stereoscopic Image Sequences

Abstract

The modeling of heads from image sequences is of great interest in the context of virtual reality, telecommunication and computerized animation systems. In this contribution a passive 3-D scanning system is described that automatically computes a complete 3-D surface model of a human head and shoulder part from a stereoscopic image sequence of a person rotating freely in front of the camera. No active element like a laser scanner is required for the modeling. 3-D shape, surface texture and the 3-D motion of the head is extracted solely from the image sequence. A dense map of 3-D surface points is obtained by image correspondence, object segmentation, interpolation, and triangulation. The surface geometry of the object is approximated by a triangular wire-frame which stores the surface texture in texture maps. Sequence processing serves to track head motion and to fuse surfaces from different view points into a consistent 3-D surface model. From the textured 3-D model highly realistic image sequences from arbitrary view points can be synthesized using computer graphics techniques.