基于主成分分析的混合象元分解

PIXEL DECOMPOSITION BASED ON PRINCIPAL COMPONENT ANALYSIS

本文主要介绍基于主成分分析的混合象元分解。通常,主成分分析用于图象增强和数据空间维数的压缩,不改变数据的空间结构。文中引入了目标检验方法,利用它可达到对混合象元的分解。由于卫片的分辨率有限,混合象元大量存在于遥感图象中,而混合象元的存在是导致异物同谱和同物异谱的一个重要原因。若能实现混合象元的分解,就可为深入应用遥感资料提供有力帮助。

In this paper,the method of mixed pixel decomposition based onPrincipal Component Analysis(PCA)is mainly introduced.Usually PCA isapplied to enhance image and reduce the dimensions of data space,withoutchanging the space construction of data.With the method of target test in thispaper,it can be realized to decompose mixed pixels.Because of the fixedresolution of Land-sat data,a large amount of mixed pixels exists in remotesensing image.Is one of the major reasons of the same spectrum for differentsurface features and different spectrums for the same surface feature.If it ispossible to decompose mixed pixels,this will be very helpful for extensivelyutilizing the remote sensing data.