光谱角度填图(SAM)是成像光谱图像处理技术中,对岩石矿物进行谱形识别的主要方法之一,但其应用效果并不十分理想。以内蒙古渣尔泰山为实验区,从岩石实验室反射光谱的相似系数聚类分析入手,研究了光谱角度填图方法的优劣,及其产生的原因。探讨了利用均值图像参与分类的处理方案,从而改善了光谱角度填图方法中易于产生“异类同谱”的不足,获得良好的岩性识别效果。

The method of imaging spectral angle mapping (SAM) is one of the main methods of mineral spectra identification in imaging spectra image processing technique, but its application result is not very good. The advantages, disadvantages and their causes are studied and discussed in this paper using clustered analysis of similarity coefficients of lithological lab spectra. The test area is located in the Zhaertai Mountain region of northern China. The imaging spectrometer data were collected using the Medium Resolution Airborne Imaging Spectrometer (MAIS) developed by Chinese Academy of Sciences. The paper emphasizes that a spectrum of rock is non linear mixture of minerals’ spectra that include either selective absorption or non selective absorption, and affection of non selective absorption to result of SAM classification. A simple average image is suggested to use in the lithology mapping technique integrated with SAM image, to solve the problem of “different objects having the same spectra” which is often produced in the classification of imaging spectra angle mapping method, thereby obtaining better lithological mapping results.