基于光谱相关角和光谱信息散度的高光谱蚀变信息提取  

Extraction of Alteration Information from Hyperspectral Imagery Based on SCA and SID

作　　者：吴浩[1] 徐元进[1] 高冉 

机构地区：[1]中国地质大学数学地质遥感地质研究所,湖北武汉430074 [2]杭州科澜信息技术有限公司,浙江杭州310000

出　　处：《地理与地理信息科学》2016年第1期44-48,F0002,共6页Geography and Geo-Information Science

基　　金：国家自然科学基金项目(41072246);中央高校基本科研业务费专项资金资助项目(CUG120116)

摘　　要：针对高光谱遥感蚀变信息提取过程中,由于混合像元的不可避免,导致蚀变矿物光谱曲线存在不同程度的失真而影响目标矿物识别精度的问题,提出一种基于光谱相关角(Spectral Correlation Angle,SCA)和光谱信息散度(Spectral Information Divergence,SID)的高光谱遥感蚀变信息提取算法(SIDSCAtan)。利用植被覆盖度表征植被信息在混合像元中所占百分比,划分出6种植被失真类型。采用不同区分方法分别比较失真光谱与理想光谱的差异,实验表明,当输入的光谱信息具有微小差异时,方法 SIDSCAtan能够做出较大的响应,在识别光谱整体形态的前提下,增强了对光谱局部特征差异的区分能力。以云南省个旧西区为研究区,运用该方法提取区内蚀变信息,应用效果较好。Hyperspectral remote sensing has been widely used in alteration information extraction.However,the spectra of altered minerals have varying degrees of distortion because of mixed pixel,which affects the accuracy of target minerals identification.A new approach to extract the alteration information from hyperspectral imagery based on SCA and SID has been presented.Using vegetation coverage to characterize the percentage of vegetation information in mixed pixel to divide the vegetation of distortion into six types.The differences between the distortion spectra and the true spectrum is distinguished by different spectral discrimination methods.The experiment shows that when the input spectral information with a slight difference,SIDSCAtan could make a big response,not only identify the whole spectrum shapes,but also enhance the partial differences of the spectral charateristics.The authors selected the west zone in Gejiu County of Yunnan Province as the study area,extracted the information of altered minerals from the hyperspectral image of the study area,the results obtained by SIDSCAtanhad higher reliability.

关 键 词：高光谱遥感 蚀变信息提取 光谱相关角 光谱信息散度 

分 类 号：P627[天文地球—地质矿产勘探]