Scattered Points Denoising of TC-Bézier Surface Fitting

Fit TC-Bezier surface with radial base function neural network, and build a radial base function network model which is suitable for surface reconstruction. A method is proposed in this paper, which suggests how to denoise and reconstructing free surface with radial base function neural network. This method simulates the internal relation between the points on surface using the study and training of scattered points made by neural cell. The result of simulation experiment shows that this model has strong capability of fitting surface, keeping the topological characteristic of the original points basically, and it also has function of smoothing noise. Besides, the speed of studying is fast and it can get surface of good fairness.