基于MNF和SVM的高光谱遥感影像分类研究

Research on Hyperspectral Remote Sensing Image Classification Based on MNF and SVM

通过分析高光谱遥感影像分类的现状及遇到的困难,以OMIS1高光谱数据为实验数据,提出了基于最小噪声分离(Minimum Noise Fraction-MNF)变换和支持向量机(Support Vector Machine-SVM)的高光谱遥感影像分类方法。分类实验结果表明:与传统的最大似然分类法(Maxi mum Likelihood Classification-MLC)比较,该方法克服了Hughes现象,分类速度得以提高,总体分类精度达到94.85%,从而表明了该方法用于高光谱遥感影像分类的实用性和优越性。

On the basis of analyzing the actuality and difficulty of Hyperspectral image classification, a method of applying Minimum Noise Fraction Transformation and Support Vector Machine to Hyperspectral remote sensing image classification is presented in this paper where OMIS 1 data is used. Compared with the traditional Maximum Likelihood Classification （MLC） method, the results show that this method overcomes the Hughes phenomenon, boosts classification speed, and has total accuracy of about 94.85 %. Thus this method demonstrated its superiority and practicability in classifying Hyperspectral remote sensing image.