基于最小二乘支持向量机的三角网格修补算法  

Hole repairing in triangular meshes based on least-squares support vector machines

作　　者：刘德平[1] 余水晶[1] 陈建军[2] 王莹莹[1] 

机构地区：[1]郑州大学机电一体化研究所,河南郑州450001 [2]西安电子科技大学机电工程学院,陕西西安710071

出　　处：《计算机集成制造系统》2009年第9期1867-1871,共5页Computer Integrated Manufacturing Systems

基　　金：国家863计划资助项目(2006AA04Z402);河南省教育厅自然科学基金资助项目(200510459060)~~

摘　　要：为实现点云数据孔洞区域的修补,提出了一种基于最小二乘支持向量机的三角网格曲面孔洞修补算法。首先检测出孔洞,采集孔洞边界周围的三角片顶点作为学习样本训练最小二乘支持向量机模型;然后对孔洞多边形进行平面填充,获得新增三角片的顶点,并用已训练好的最小二乘支持向量机模型将其优化,最终实现孔洞的修补。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,为孔洞修补研究提供了一种新思路。To realize hole repairing of cloud data, a novel algorithm for repairing the holes in triangular mesh surface based on Least-Squares Support Vector Machines （LS-SVM） was put forward. Firstly, the holes were identified and the LS-SVM model was trained by using the triangle vertices which gathered around the holes. Secondly, the new triangle vertices were obtained by triangulating the characteristic polygons of the hole. Finally, coordinated values of the new triangle vertices were optimized by the LS-SVM mode and the holes were repaired. Conclusions of experiments indicated that this method had higher precision and faster processing speed than Artificial Neural Network （ANN）. It was proved to be a practical and new way for holes repairing.

关 键 词：逆向工程 最小二乘支持向量机 孔洞修补 三角网格曲面 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]