基于SVM的决策树多类分类器及在遥感图像中的应用

摘要

利用三种不同的聚类方法——利用类均值的最小距离聚类、利用类均值的最大距离聚类和利用最大间隔准则聚类，提出了三个基于SVM的决策树多类分类器．为了检验所提算法的有效性和先进性，对AVIRIS遥感图像进行了实验．实验结果表明，本文所提的三种算法明显好于最小距离分类法、线性判别分类法、决策树分类法、OAR—SVM和0AO-SVM．%In this paper, three kinds of decision tree multi-class classifiers based on SVM are pres- ented by means of three clustering methods, which are respectively clustering with minimum distance of class means, maximum distance of class means and maximum margin criteria. The experiments with AVIRIS remote sensing image are made for testing the validity and advantage of our proposed algo- rithms. The experimental results demonstrate that our methods are significantly better than minimum distance classification, linear discriminant classification, decision tree classification, OAR-SVM and OAO-SVM.