Surface reconstruction from sliced point cloud data for designing facial prosthesis

People with facial deformities often use prosthetic devices to restore their normal visual appearance. A prosthesis device is a custom-sculpted artificial facial feature made of silicon that is worn to cover deformity and restore form and hopefully function of the feature. Surface scanning, computer-aided design, and rapid prototyping technologies are being used to create and evaluate these customized prostheses. The geometry of the patient's existing facial features can be captured in few seconds using non-contact 3D scanners. However, the resultant point data set is very large and corrupted and should be reduced to reconstruct accurate facial surface. It is also equally important to evaluate the fitting characteristics of designed prosthetic device on patient's face for realistic effects. In this paper, an integrated contour-based algorithm is presented, that facilitates a unified approach for data simplification and surface modeling of face and prosthetic device. For data simplification, an ordered sequence of contours from a dense unstructured point cloud data is extracted and simplified as B-spline curves with a reduced number of control points. The simplified-extracted contours are then used to generate the facial lofted surface. This facial model is then used for modeling and fitting evaluation of the prosthetic device. A prosthetic device is designed using contour matching of a generic feature model and the patient's facial model to ensure its fitting accuracy on patient's facial features. Facial data with a simulated nasal deformity is used to illustrate this methodology. The effectiveness of the method can be improved by increasing the number of parallel contours, and using a larger second stage reduction ratio during data simplification.