This paper proposes an efficient method to reconstruct human face models from 3D face scans using an adaptation algorithm based on regional information. The resulting model can represent individual facial shapes with various expressions, establishing dense correspondences across the whole facial expression sequences. We first initialize the global shape of the face model by implementing RBF interpolation. An efficient vertex mapping is then developed to reconstruct the surface details. The new 3D face are finally created by remapping the scan texture to the facial surface. Experiments show that our method can create effective 3D face models for continuous and realistic 3D facial animation.