为了实现对无任何先验知识的高光谱遥感数据的全自动分类，提出了一种关于高光谱图像的无监督分类算法。该算法将高光谱图像的凸面几何特征与光谱特征相结合，通过自动提取端元，并利用所提取的端元进行类别识别来实现高光谱图像的自动分类。此算法的特点是原理简单、易于实现、适应性广，而且不需要任何辅助支持和人工干预。实验结果表明，该算法能够获得较好的分类效果。

In order to classify the data of Hyperspectral remote sensing images automatically without prior knowledge, an unsupervised classification algorithm is presented based on the conception of convex geometry and spectral features in this paper. The endmembers are selected step by step during processing and each endmember can be identified as one class. The advantages of this algorithm are simple in theory, easy to accomplish, widely used, and without any manual assistance. The experiment shows that the classifying result of this algorithm is satisfied.