Based On Neural Network Technology Research Of Free-Form Surface Reconstruction

With the development of reverse engineering, free-form surface reconstruction is increasingly important in many fields such as CAD/CAM. Now, with the development of neural networks (ANN) theory. surface reconstruction of neural networks has been put forward at home and abroad, putting forward free surface modeling method of data of scattered point cloud from a new angle, because of its advantages of fast modeling speed, high modeling accuracy, good smoothness and fine noise immunity, domestic and foreign experts have conducted a lot of research on it.This paper research on BP, RBF and SOFM neural network deeply, by lots of modeling and calculation completing noise point removed, data recovery, region segmentation and surface reconstruction of free-form surface which based on neural network. And making numerous simulation on the technique of free-form surface reconstruction which based on neural network by MATLAB, the results of engineering examples showed that:reconstructed surface which uses the technology of BP neural network de-noising technology, data recovery technology RBF neural network and BP network reconfiguration not only increases the accuracy than traditional surface reconstruction by2times, but also increase reconstruction of efficiency by50%, this provides a certain theoretical significance and practical value.