Surface reconstruction from cloud points based on support vector machine

Abstract

Surface reconstruction based on Support Vector Machine (SVM) is a hot topic in the field of 3D surface construction. SVM based method for surface reconstruction can reduce the noise in the sampled data as well as repair the holes. However, the regress speed of SVM is too slow to reconstruct surface quickly from cloud points data set which has a lot of points. In this paper, a feature-preserved nonuniform simplification method for cloud points is presented, using which to simplifying points set. This method simplifies the data set to remove the redundancy while keeping down the features of the model. Then the surface is reconstructed from the simplified data using SVM. Both theoretical analysis and experimental results show that after the simplification, the performance of method for surface reconstruction based on SVM is improved greatly as well as the details of the surface are preserved well. © 2008 IEEE.