基于SVM和空间投影的点云空洞修补方法.pdfVIP

第35 卷 第22 期 计 算 机 工 程 2009 年11 月 Vol.35 No.22 Computer Engineering November 2009 ·开发研究与设计技术· 文章编号：1000—3428(2009)22—0269—03 文献标识码：A 中图分类号：TP391 基于SVM 和空间投影的点云空洞修补方法 蒋 刚 (西南科技大学制造科学与工程学院，绵阳 621010) 摘 要：为实现在逆向工程中的点云空洞修复，在理论研究的基础上，通过空间投影获得二维数据，采用支持向量机做回归分析，获得残 缺点的坐标参数，从而完成空洞修补，运用数字实验对该方法的可行性进行验证，仿真实验结果表明，该方法可以获得良好的修补效果， 能够为曲面建模和数控系统G 代码的生成提供完整的点云数据，具有一定应用价值。 关键词：支持向量机；点云；空洞修补；空间投影 Point Cloud Hole Filling Method Based on SVM and Space Projection JIANG Gang (School of Manufacturing Science Engineering, Southwest University of Science Technology, Mianyang 621010) 【Abstract 】In order to implement the point cloud hole filling in reverse engineering, this paper adopts space projection technology and Supported Vector Machine(SVM) based on Statistic Learning Theory(SLT). Coordinate value of fragmentary points can be gotten by SVM regression, and the whole filling work can be finished by this method. Some numerical experiments have been applied to verify the feasibility of this method. Simulation experimental results show its good performance, and it can give complete points for curve modeling and G code generation in the next step. It has the application values. 【Key words 】Supported Vector Machine(SVM); point cloud; hole filling; space projection 1 概述 min 1 || w ||2 +C n (ξ +ξ ∗ )