Deforming shape towards target for 3D face recognition

Expression is the key issue for 3D face recognition. A novel deformation matching approach is proposed in this paper. First, a guidance-based deformation is applied to the probe to strengthen its discriminative characteristics. We use radial basis function to fulfil the nonrigid registration that sets up point set correspondence and fits model deformation. This nonrigid registration is used to produce the deformed model which used to surface matching with the target in the second step. Extensive experiments on FRGC v2 databases illustrate that the proposed algorithm outperforms the rigid matching method. © 2011 by Binary Information Press.