Stereoscopy based 3D face recognition system

AbstractThis paper presents an innovative facial recognition algorithm: a stereoscopic calibrated system acquires two images of a subject that are then separately analyzed by an active appearance model code which extracts from each of them 58 homologous points. The triangulation of these points allows the construction of a three-dimensional mask of the face of the individual. This mask is then compared against a database of reference masks. The obtained results are encouraging: preliminary tests on a limited database showed a false rejection rate identically equal to zero and a false acceptance rate of about 3%