基于组合降维及旋转SVM的高光谱图像分类  

Hyperspectral image classification based on combined dimension reduction and rotation SVM

作　　者：任荣 保文星[2] REN Rong;BAO Wenxing(School of Computer and Information, Hefei University of Technology, Hefei 230009, China;School of Computer Science and Engineering, Beifang University of Nationalities, Yinchuan 750021, China)

机构地区：[1]合肥工业大学计算机与信息学院,安徽合肥230009 [2]北方民族大学计算机科学与工程学院,宁夏银川750021

出　　处：《合肥工业大学学报（自然科学版）》2018年第12期1617-1621,1651,共6页Journal of Hefei University of Technology：Natural Science

基　　金：国家自然科学基金资助项目(61461003)

摘　　要：高光谱图像的超多波段可以描述丰富的地物信息,但是也带来了维数灾难的问题。文章提出了主成分分析(principal component analysis,PCA)与线性判别式分析(linear discriminant analysis,LDA)组合降维方法,使类内距离最小化、类间距离最大化,有效地消除数据冗余并保留主要信息量,保证了降维后的数据具有最佳区分度;旋转森林是一种先进高效的集成学习算法,将基分类器由决策树改进为支持向量机(support vector machine,SVM),并将组合降维后的数据应用于旋转SVM分类器,分类精度有了显著的提高。实验对比分析表明,该方法具有较好的分类效果。Super multi-bands of hyperspectral image can describe abundant object information,but it leads to the curse of dimensionality.A combined dimension reduction method based on principal component analysis(PCA)and linear discriminant analysis(LDA)is proposed.The distance of inner classes is minimized,and the distance of inter classes is maximized.Data redundancy is eliminated effectively,and the main information is retained.This method ensures the dimension reduction data set has the best discrimination.Rotation forest is an advanced and efficient ensemble learning algorithm.The decision tree base classifier is replaced by support vector machine(SVM).The data set after combined dimension reduction is applied to the rotation SVM classifier,and the classification accuracy is improved significantly.The experimental analysis shows that the proposed method achieves better classification results.

关 键 词：高光谱图像 分类 集成学习 支持向量机(SVM)分类器 降维 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]