Cranio-facial Reconstruction Based On Landmarks Deformation

3D cranio-facial reconstruction (CFR) is to reconstruct the 3D facial model from the skull data based on the relationship between skull and facial surface, which is a challenging research topic in computer graphics. CFR can be widely used many applications, such as criminalistics, archeology, face animation, face-lifting and so on. In the paper, we firstly summarized the state-of-the-arts of CFR, and we then present a new approach of CFR based on landmarks deformation. Experiments show the high efficiency of our approach.Some deformation methods of CFR usually need a reference skeletal model from a model database. In this paper, we present a CFR method which doesn't require the reference. We investigate the relationship between the predefined landmarks on the skull data and facial model, and the deformation algorithm of Radial Basis Function (RBF) is used to generate the 3D face model. In addition, the determination of facial landmarks plays an important role in the similarity of deformation facial model. The popular methods for the facial landmarks are to use the tissue thickness on the reference skeletal surface. We also propose the semiautomatic reconstruction method based on the implicit surface of reference facial model. The experimental results show the high efficiency of our method. We finally develop a prototype system of computer-aided facial reconstruction--Face3DRecon to validate the algorithms we presented.Moreover, we also employ some interpolation reconstruction methods purely depended on the landmarks, and we find that the RBF interpolation method for CFR is superior to the other methods. The deformation approach we presented has more advantages in the further processing such as facial expression, face-lifting and texture mapping and so on.