A facial authentication system using complex-valued eigenfaces

In our previous papers at the SICE2003, we proposed the complex-valued eigenface method for the efficient representation of 3D normal vector maps and its application to face recognition. But the uncertainty of pose and position of the face can significantly decrease their performance. In this paper, we propose an automatic method for adjusting pose and position by placing one axis and several control points on the face based on the symmetry and Gaussian curvature of the 3D profile and intensity image.