CHRIS 高光谱图像森林类型分类方法比较研究

以长白山为试验区,选择CHRIS/PROBA高光谱零度角遥感数据,在对其进行预处理的基础上,通过应用最大似然法(MLC)、最小距离法、支持向量机法(SVM)和光谱角填图法(SAM)等几种常用的高光谱遥感分类方法对影像进行森林类型分类。利用混淆矩阵对分类结果进行验证,结果显示:在高光谱遥感森林类型分类中,SVM总体分类精度最高,为84.60％;其次是MLC,为 83.53％,最小距离法73.81％,SAM 56.49％。Kappa系数从高到底为:SVM 0.78,MLC 0.77,最小距离法0.68,SAM 0.52。经过比较分析,得出SVM分类方法精度最高,这表明该方法用于高光谱遥感森林分类中的实用性和优越性。

Comparation of Forest Types Classification Methods Using CHRIS Hyperspectral Image

The Changbai Mountain was regard as experimental area in this text, based on the hyperspectral remote sensing data of CHRIS/PROBA O degrees was selected and preprocessed, several classification means of hyperspectral remote sensing to forest types classification of image were used ,such as maximum likelihood method (MLC), minimum distance method. Support vector machine (SVM) method and Spectral Angle mapping method(SAM) etc. Finally,the real reference sources were used to verify the classification results. The results showed that SVM got the highest accuracy of 84. 60% among all the forest type classi-fication methods, the accuracies followed were MLC (83.53 %), minimum distance method (73.81%) and SAM (56.49%). The Kappa coefficients were displayed from high to low: SVM (0. 78), MLC (0.77),minimum distance method (0. 68) and SAM (0. 52). After the comparison of classification results, SVM obtained the highest accuracy in all classification methods. It showed the practicability and advantage of SVM applied to forest classification.