基于光谱特征空间的苏必利尔湖型铁矿高光谱遥感蚀变信息提取方法  

The Application of Spectral Characteristic Space Method to the Alteration Information Extraction of Lake Superior-Type Iron Deposit

作　　者：黄爽[1] 陈圣波[1] 周超[1] 査逢丽 路鹏[1] 黄唯实 

机构地区：[1]吉林大学地球探测科学与技术学院,吉林长春130026 [2]辽宁省第七地质大队,辽宁丹东118000

出　　处：《地球科学（中国地质大学学报）》2015年第8期1325-1329,共5页Earth Science-Journal of China University of Geosciences

基　　金：国家高技术研究发展计划(863计划)项目(Nos.2012AA12A308;2008AA121100);国家自然科学基金项目(No.41402293);吉林大学研究生创新基金资助项目(No.2014029)

摘　　要：由于境外铁矿床位置偏远,现场研究困难,旨在探讨苏必利尔湖型铁矿的高光谱遥感蚀变信息提取方法,为境外找矿提供依据.鉴于传统蚀变信息提取方法的局限性,研究了在二维光谱特征空间中提取蚀变信息的方法.高光谱影像经主成分变换后,包含较多地物信息的两个主成分的二维光谱特征空间体现了各向异性的特点,不同地物的聚类常呈类似椭圆分布.在蚀变信息的聚类椭圆中圈定散点,并应用波谱特征拟合,将异常散点的平均光谱与美国地质调查局(United States Geological Survey,USGS)光谱库中的矿物光谱匹配,确定蚀变矿物类型.以苏必利尔湖型铁矿——巴西Aguas Claras铁矿区为例,提取了赤铁矿、绿泥石等蚀变矿物,结果表明矿物的类型和分布与该区地质概况吻合.The field study of overseas iron ore deposits is difficult because of its remote location. The spectral characteristic was used to extract alteration information in view of the limitation of the traditional methods in this paper, aiming to explore altera- tion information extraction from hyperspectral images for the Lake Superior type iron ore to facilitate overseas prospecting. A two-dimensional spectral characteristic space takes on an anisotropic feature in associated distribution of two principal bands af- ter principal component analysis. The distribution is usually combined by oval clusters. The scatter points were enclosed in oval clusters of alteration information, and the mean spectra of abnormal scatter points were matched with mineral spectra from United States Geological Survey（USGS） spectral library by means of spectral feature fitting to ensure the type of alteration minerals. Spectral characteristic space method was described with the instance of Lake Superior-type iron deposit--Aguas Claras iron area in Brazil. Hematite, chlorite and other minerals were extracted in this study. It is found that the type and distribution of extracted altered minerals were consistent well with geological condition.

关 键 词：光谱特征空间 蚀变矿物 二维散点图 遥感 Hyperion数据. 

分 类 号：P627[天文地球—地质矿产勘探]