岩石实验室光谱对应分析

Correspondence Analysis of Laboratory Spectra of Rock

成像光谱图像具有几十个窄波段和高的光谱分辨率，使直接识别岩石岩性成为可能，但其海量数据又为数据的处理、分析和应用带来了不便。因此就需要寻求一种在不损失信息或最大限度保持原有信息的前提下又压缩或最小化数据量的方法。在对岩石实验室光谱按成像光谱仪（ＭＡＩＳ）通道重采样后，应用了一种特征信息提取的方法（对应分析法），对岩石光谱进行了分析，达到了压缩数据与岩石分类的目的。

Imaging spectrometer has tens or hundreds of bands. Th us, it provides the opportunity to acquire more volume of information about the surface targets. At the same time it is difficult to process so huge volume of d ata, thus, it is important to find a solution to this problem. Principal Compone nt Analysis (PCA) is a good method to reduce the volume of data at the cost of a little loss of information. But one of disadvantages of PCA is that each PCA tr ansformed component has a blur meaning. Correspondence Analysis (CA) overcomes t his shortcoming. This paper uses CA to process the Laboratory spectra of nine ro cks, which have been resampled on the channels of MAIS, and get a good result. It is also a meaningful attempt to the processing of imaging spectrometer data.