Level of detail control of a virtual dental training system

This paper presents a level of detail control method for a virtual dental training system, the input of which is a triangulated surface model without the property of continuity commonly satisfied in other methods. This method uses relative curvatures of vertices as the criteria for multiresolution modeling. The primary strategy of this method is to produce tooth models with needed level of detail by vertex decimation. After illustrating our motivation and consideration to develop this method, the paper introduces the policy for vertex decimation and the computation method of relative curvature at first. Then the level of detail control method is presented in detail. And we adopt both maximum and average error metrics to measure the geometric fidelity between the original model and approximations. Finally, some sampled results are illustrated and analyzed in depth. The experiments demonstrate that the method can provide good level of detail control for the tooth models with reasonable time cost and geometric deviations.