基于遥感影像的最大似然分类算法的探讨

A Study of Maximum Likelihood Classification Algorithm Based on Remote Sensing Image

通过对最大似然分类算法的研究,并将其与马氏距离分类法在相同情况下进行对比试验,结果表明,最大似然分类方法对于光谱特性呈正态分布的遥感影像能提供较高的分类精度,而对于光谱特性呈非正态分布或偏离正态分布总体的遥感影像,最大似然分类法的实际分类效果并不理想,进而分析最大似然分类法的适用性.

In this thesis, the research of maximum likelihood classification algorithm attaching with Mahalanobis distance classification with the same circumstances in the comparative test results show that the maximum likelihood classification method can provide higher classifiable precision for those remote sensing images with normal distribution on spectral characteristics, while it is actually not idealistic for those abnormal distribution ones or deviating from a normal system. Then the adequacy of the maximum likelihood classification method will explore.