Photo-realistic 3D human face modeling based on radial-basis function networks

This paper describes an efficient method for modeling photo-realistic 3D human face from two orthogonal photos taken from front and side views. The method utilizes active shape model based on RGB color space to extract facial boundaries and detect feature points from the orthogonal photos. The previously constructed generic models are modified by exploiting the abilities of radial-basis function networks and considering the feature points detected from the orthogonal photos. Based on image mosaic, the Laplacian pyramid is developed to obtain 360 degree seamless texture, in the manner of feature fusion. Simulation results show that the proposed method outperforms the conventional methods with respect to the modeling precision, and that the 3D faces generated by the proposed method are both smoothing and photo-realistic.