联合光谱角匹配与随机森林的蚀变信息提取  

Extraction of Alteration Minerals Information Based on SAM and RF

作　　者：赵杰[1,2,3] 周可法 崔世超[1,2,3] 王金林[1,2] ZHAO Jie;ZHOU Kefa;CUI Shichao;WANG Jinlin(Xinjiang Research Center for Mineral Resources,Xinjiang Institute of Ecology and Geography Chinese Academy of Sciences,Urumqi 830011,China;Xinjiang Institute of Ecology and Geography,Chinese Academy of Sciences,Urumqi 830011,China;University of Chinese Academy of Sciences,Beijing 100049,China)

机构地区：[1]中国科学院新疆生态与地理研究所新疆矿产资源研究中心,乌鲁木齐830011 [2]中国科学院新疆生态与地理研究所,乌鲁木齐830011 [3]中国科学院大学,北京100049

出　　处：《遥感信息》2017年第6期51-55,共5页Remote Sensing Information

基　　金：新疆维吾尔自治区重大专项(201330121-2);中国科学院院地合作项目(Y437051)

摘　　要：蚀变矿物信息提取是高光谱遥感在地质方面的重要应用,如何快速精确地进行蚀变矿物信息提取一直是一个难点。随机森林是近年来发展起来的一种新算法,在分类与回归方面具有优异的表现。但蚀变矿物信息提取不同于一般的遥感分类,针对这一算法在高光谱蚀变矿物信息提取中的适用性问题,提出了一种将随机森林算法与光谱角匹配技术相结合的提取方法。实验中通过光谱角匹配选取训练样本,然后构建随机森林分类模型,并利用美国内华达州Cuprite铜矿区的AVIRIS高光谱影像提取了明矾石与高岭石2种矿物的分布分范围,获得了较好的效果。Alteration minerals information extraction is an important application of hyper spectral remote sensing in geology.But it is a difficult problem to extract alteration minerals information quickly and accurately.Random forest(RF)is a new algorithm developed in recent years.It has excellent performance in classification and regression.However,the alteration minerals information extraction is different from the general classification of remote sensing.In view of the applicability of this algorithm,a novel method is proposed by combining random forest algorithm with spectral angle mapper(SAM).In the experiment,we select training samples by spectral angle matching,then construct a random forest classification model,and carry out the extraction experiment of alunite and kaolinite using the airborne visible/infrared imaging spectrometer(AVIRIS)data at Cuprite,Nevada.The results of the study show that this method is valuable for identification of alteration information.

关 键 词：随机森林 光谱角匹配 高光谱遥感 蚀变矿物信息提取 集成学习 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置]