Craniofacial model reconstruction from skull data based on feature points

3D craniofacial model reconstruction from skull data has been a challenging research topic for many years. In this paper we develop a craniofacial reconstruction system based on feature points, and our approach only requires the skull model from a standard 3D scanner and feature points superimposed on the skull data. Our method consists of three stages: hole repairing for the 3D skull model, facial reconstruction with Radial Basis Function(RBF) deformation and texture mapping. A new advancing layer-wise solution algorithm and a template matching method are proposed for repairing big and particular holes on the skull model. The reference face model is then transformed to reconstruct the face model base on marked feature points on the skull model, and we improve the RBF deformation.. The experiments demonstrate the high efficiency and visual realism achieved in our approach.