支撑向量机在高光谱遥感图像分类中的应用  

Application of Support Vector Machines in the Classification of Hyper-spectral Remote Sensing Image

作　　者：许将军[1] 赵辉[2] 

机构地区：[1]中国民航飞行学院广汉分院,四川广汉618307 [2]电子科技大学自动化工程学院,四川成都610054

出　　处：《计算机仿真》2009年第12期164-167,共4页Computer Simulation

摘　　要：高光谱遥感图像具有维数高的特点,当样本较少时,利用传统的统计识别方法分类,分类精度低。可支撑向量机(SVM)能解决小样本、高维、非线性分类问题。采用归一化法对原始图像做预处理,再分析不同的SVM核函数对分类精度的影响;并把SVM与最小距离法,马氏距离法等的分类结果进行比较。结果表明SVM的核函数类型对分类正确率影响不大,其分类精度高于传统的统计识别方法。The characteristic of hyper- spectral remote sensing data is high dimensional. The accuracy of traditional classification methods is always unsatisfactory. Support vector machine （SVM） is a good method for solving limited sample, high dimensional, and nom -linear classification problem. The image data are preprocessed by Image normalization method, the influences of different kernel functions on classification accuracy are analyzed, and the classification results by SVM are compared with that of other methods （ such as Minimum Distance method, Mahalanobis Distance Method）. Experiment results show that the classification accuracy is almost identical for different kernel functions, and VM method has better classification and recognition accuracy than traditional algorithms.

关 键 词：高光谱遥感 图像预处理 支撑向量机 核函数 

分 类 号：TF75[冶金工程—钢铁冶金]