摘要
简要介绍了高光谱遥感发展的现状,讲述了EM算法的工作原理及其在高光谱影像分类中的具体计算方法,对于噪声,采用一定的方法加以判断并予以剔除,对余下的未标识样本点给与适当的权值,并应用于EM算法的迭代过程中。通过实验验证,经过去噪处理的EM算法可以有效地改善高光谱影像分类结果。
The paper first introduced the development of hyperspectral remote sensing, then discussed the fundamental of EM algorithm and the definite calculating process in hyperspectral classification, finally the author proposed the method to deal with noise, including the estimation of the noise and assigning appropriate weights to the unlabeled samples. The experiments demonstrated that the ameliorated EM algorithm can effectively enhance the hyperspectral classification performance.
出处
《海洋测绘》
2005年第6期18-21,共4页
Hydrographic Surveying and Charting