Pose-varied face recognition based on 3-D face model
Guohui He, Bin Zhu, Junying Gan · 2008
In this paper, 3D face model is utilized to tackle position problems in face recognition. At first, RBFS (radial basis function) and B curve method are combined to carry on 3D face reconstruction. Then face pose parameters can be estimated rapidly when three angles of facial rotation are calculated separately. Later, face position parameters and 3D face model are combined to generate new face image whose position parameters are the same as the input ones. At last, 2D face recognition is done with the same face positions. Experiments demonstrate that the method can get good recognition results, can be simply operated and is a valid face recognition method used for solving face position problems.