Mesh Simplification Algorithm And Improved

Operations on 3D models are implemented in many areas. Models generated by CAD software or high resolution scanners are usually over dense and contain massive excess data, thus not suitable for interactive operation or even unsustainable for lower work stations. To accommodate this, methods that can both simplify the model and retain much of its original visual information are proposed, based on different trade-off between time\space consumption and visual quality preservation. This article traces the development of these algorithms along with their idea of model data reduction and compromise between simplification time and output quality, after on enough analysis about know methods, a refined one is presented who try to trade the depth of simplification algorithm for computation time and output quality.The basic principle of the algorithm design is:while the simplification doesn't go too deep, outputs produced by cheap election are not significantly worse than those produced by expensive ones, so cheap ones can be used to reduce time, and adopt lower threshold at the same time to retain output quality, but in the expense of ignoring some of the in fact safe operation due to the inaccuracy of cheap election, that is, trade depth of simplification for time and quality. The improved method uses classical edge collapse mechanism to reduce model data, edges are elected by cheap meaning based on similarity of surface normal, and adopt very small dissimilarity tolerance to prevent quality losses. New method doesn't queue candidate edge via their safety measure and try to transmit or approximate all related high time-consuming mathematical functions in order to save processing time further more.Experiments show that refined method can reduce original model size to its 50%-25% using significantly shorter time than popular ones based on QEM election, and the outputted simplified model can retain original model fairly well. Another advantage of the new method is its sensitivity to the regularization of the input model. Input with relatively smoother surface requires much less time to reduce to 25% using new method. But if the model is reduced to 10% or lower, the output would loss shape compared to the original model, means this method is not suitable for very deep simplification. While the depth of simplification is not demanding, new method left much greater surplus processing time than popular methods but a little bit low on output quality, adjustment should be made to trade time for quality, besides, generalizing the method to accommodate models with non-geometry field and arbitrary topology is major future work.