成象光谱遥感图象的有限光谱混合分析

Spectral Mixture Analysis of Imaging Spectrometer Data

本文提出了一种成象光谱数据处理的新方法──有限光谱混合分析方法ＣＳＭＡ（ＣｏｎｓｔｒａｉｎｅｄＳｐｅｃｔｒａｌＭｉｘｔｕｒｅＡｎａｌｙｓｉｓ），该方法将图象中每一象素看作多种类别地物的混合，利用成象光谱仪各波段反射值计算各个类别所占的比例，这是一个求解混合光谱方程最优解的过程。本文提出的有限光谱混合分析方法是在约束条件下求混合光谱方程的最优解，并提出了快速算法。该方法可以得到比传统遥感分类更有意义的结果，使遥感技术更趋于定量化、实用化。

Imaging Spectrometer is a new developed remote sensing instrument. A newprocessing method of its data-Constrained Spectral Mixture Analysis (CSMA) is presented in this paper. CSMA looks every pixel as mixture of some endmembers. Thecontent of each endmember in a pixel can be calculated using Imaging Spectrometerdata. This is a problem of getting optimal solution of the spectral mixture equation.CSMA gets the optimal solution of the equation with some conditions and a fastalgorithm is proposed here. This result is more interesting than traditional classificationbecause it is more quantitative.