A New Method to Classify Hyperspectral Data of Landsat TM Image

摘要

Classification, as well as finding representative spectrum, is always one of the major applications of satellite hyperspectral image data. A new method to find the set of representative spectrums was introduced in this study. It was firstly used to classify forest types in Tianmu Mountain National Nature Reserve by using Landsat TM images. This method is generic and can be applied to the data set that is hard to find one representative data for one class.