Face 3D Modeling Based on Projective Rectification

Abstract

Face 3D modeling is the difficulty problem in the field of computer graphics, computer vision and artificial intelligence. In recent years, it has become the most active research focus both at home and broad. 3D modeling of face is the key step to realize face recognition, and the technique of face 3D modeling has obtained extensive applications in many fields, such as film, animation, interactive games, video conference, human-computer interaction, reverse engineering, medical and public safety. In this paper, the technology of face 3D modeling based on projective rectification is presented and the reconstruction of face 3D digital model can be achieved by it. Firstly, BP neural network is used to simulate the mapping relationship between the 3D object and its images. Then, the rectification on the left and right images acquired by stereo vision system is implemented according to the principle of epipolar line constraint. On the left and right rectified image planes, the match researching of corresponding points are reduced from 2D plane to the horizontal lines, so the image matching and face 3D modeling can be implemented efficiently.