SVM加权学习下的机载LiDAR数据多元分类研究  

Aerial LiDAR Data Classification Using Weighted Support Vector Machines

作　　者：吴军[1] 刘荣[1] 郭宁[1] 刘丽娟[1] 

机构地区：[1]桂林电子科技大学电子工程与自动化学院,桂林市金鸡路1号541004

出　　处：《武汉大学学报（信息科学版）》2013年第1期1-5,共5页Geomatics and Information Science of Wuhan University

基　　金：国家自然科学基金资助项目(60962003);广西研究生教育创新基金资助项目(YCSZ2012074)

摘　　要：基于支持向量机统计学习分类过程中不同特征对分类结果贡献存在差异的问题,提出了支持向量机加权学习下的训练、分类新方法,以实现对城区机载LiDAR数据多元分类(地面、树木、建筑),并对特征矢量加权归一化、特征权重计算以及该方式下多元分类策略的建立进行了讨论,实验证明了该方法的有效性。This paper presents our research on classifying scattered 3D aerial LiDAR height data into ground,vegetable（trees） and man-made object（buildings） using improved Support Vector Machine algorithm.To this end,the most basic theory of SVM is first outlined and with the fact that features are differed in their contribution to identify certain class or classes simultaneously,Weighted Support Vector Machine（W-SVM） technique is developed for maximizing the ＂recognition＂ capacity of SVM features in classifying scattered 3D LiDAR height data.Second,we give a proof that the implement of W-SVM is equal to the features normalization multiplied by one weight that indicates feature＇s contribution to certain class or multi-class as a whole.The weight calculation for each feature is discussed as well.Third,Based on W-SVM technique,one 1AAA1 solution to multi-class classification is proposed by integration ＂one against one＂ and ＂one against all＂ solution together.Finally,the experiment of classifying LiDAR data with presented technique is presented and shows encouraging improvement classification accuracy,compared to tradition SVM technique.Valuable conclusions are given as well.

关 键 词：LiDAR分类 支持向量机 监督分类 

分 类 号：P237.4[天文地球—摄影测量与遥感]