中文摘要：

分析了点云建模的特点,将基于统计学习理论的支持向量机引入该领域。首先提取点云数据中的强特征,采用支持向量回归机构建轮廓;然后在轮廓形成的不同区域分别提取弱特征,用回归的方式逐步重构区域纹理,从而得到整个物体的表面表达。理论分析和实验结果表明该方法的精度和处理速度优于人工神经网络,具有一定的实用性,为点云建模研究提供了一种新的思路。

英文摘要：

Via the analysis to the features of modeling from points, a statistical leaning method was proposed in this field.. This method is based on Support Vector Machine （SVM）, which is state of the art in machine learning. Regression curves were generated by SVM after distilling features from point clouds. Skeleton was formed by robust features and region grid came from feeble features step by step, and then, the whole object surface grid was built up. Conclusions can be made by theory analysis and experiment results that SVM based method has higher precision and faster processing speed than Artificial Neural Network （ANN）. It is a practical and new way for modeling from points.