Extraction of semantic 3D models of human faces from stereoscopic image sequences
Information and Communication Theory Group Faculty of Information Technology and Systems, Delft University of Technology Mekelweg 4, 2625 CD Delft, The Netherlands
Information and Communication Theory Group Faculty of Information Technology and Systems, Delft University of Technology Mekelweg 4, 2625 CD Delft, The Netherlands
Information and Communication Theory Group Faculty of Information Technology and Systems, Delft University of Technology Mekelweg 4, 2625 CD Delft, The Netherlands
Information and Communication Theory Group Faculty of Information Technology and Systems, Delft University of Technology Mekelweg 4, 2625 CD Delft, The Netherlands
Information and Communication Theory Group Faculty of Information Technology and Systems, Delft University of Technology Mekelweg 4, 2625 CD Delft, The Netherlands
Abstract
In this paper a system is described for the automatic generation ofa specific semantic 3D model of a human face from a recorded stereoscopic image sequence. In the acquisition stage a calibrated stereoscopic camera records the specific face. After rectification of the image sequences a dense disparity field is estimated using a Markov Random Field approach. Next a triangle based 3D-wire frame model of the specific face is reconstructed by triangulation of corresponding left and right pixels. After compensation for rotation and translation, a triangle based 3D-wire frame model of a generic face is elastically matched to the specific face. As a consequence of this approach the resulting 3D model accurately represents the geometry of the subject with the availability of the face semantic information which is required for realistic representation and animation of human faces as well as for accurate facial expressions estimation.