3D face recognition

Abstract

This report presents a face recognition system based on 2D recognition methods and 3D face data. 2 sets of stereo cameras are used to acquire the 2D and 3D face data of a human subject. By applying 2D face detection on the images, the face was first located in the 2D domain. After mapping the 2D location to 3D, 3D face points cloud are generated from each stereo camera. Then the two 3D cloud data are merged together. The orientation of the face is approximated by the relative position of the face to the two stereo cameras, and a frontal face is generated based on this approximation. And at last face recognition methods are then applied on the generated 2D image. Experimental results involving 10 persons with different facial expressions and extracted from different viewpoints have demonstrated the efficiency of our algorithm.