Abstract
In forensic science and archeology, facial reconstruction means the techniques to reconstruct the face of a man from his cadaver. Because of the massive time and labor-intensive processing of the traditional methods, computer graphic techniques have been rapidly developed in recent years to reduce the processing time and to enhance the performance of reconstruction. In this study, we process the facial reconstruction by adopting an elastic model. The energy generated by the displacement of each point in the model would cause the points in the neighborhood to have the corresponding displacement in order to reduce the total potential energy of the model. The desired result of facial reconstruction is accomplished when the minimum value of the total potential energy of the point set is achieved. A 3D model of a skull is created from the CT images of the cadaver to be classified. We then choose the best match of the skull model in the database with the skull model of the cadaver as the reference model. The result of the registration between the reference model which includes hard and soft tissues register and skull model of the cadaver is set to be the initial position of the deformation that follows. The goal of deformation is to reduce the total potential energy of the elastic model so that the point set of the model will be finally distributed in the form that achieves the minimum value of the total potential energy. In addition, a weighting parameter is considered to refine the model.