基于主成分分析的SVM矿化信息提取研究——以青海黄南州阿哇地区为例  

Extracting Alteration Information by SVM Based on Principal Component Analysis

作　　者：薛云[1] 戴塔根[1] 邹艳红[1] 杨自安[2] 邹林[2] 

机构地区：[1]中南大学地学院,长沙410083 [2]有色金属矿产地质调查中心,北京100012

出　　处：《遥感信息》2007年第6期32-35,44,I0004,共6页Remote Sensing Information

基　　金：中国地质调查局地质大调查项目(1212010660601)

摘　　要：针对传统矿化信息提取方法需要大量样本,且样本选取困难的缺陷,利用主成分分析和支持向量机(SVM)原理,建立矿化信息提取模型。选择青海黄南州阿哇地区作为典型研究区。首先进行主成分分析,选取训练样本;然后求解最优超平面,进而确定决策函数;最后泛化推广识别其他待识别的样本。通过所提取的遥感蚀变异常信息与原有矿区叠加分析,叠加基本吻合;从野外实地验证来看,均发现了不同程度的矿化现象,并指出了5个重点异常区。In the past, many samples were demanded and it is difficult that high quality samples were selected. A new method for extracting mineralization from remote sensing image by SVM based on PCA is presented in the paper. The research was done in Awa area, Huangnan city. Firstly, training sample was selected by PCA; Secondly, the most optimal hyperplane and decision-making function were found; At the end, alteration information was extracted. Through on the spot investigation and comparing with data of the known alteration areas in the mineral geology mapping of field, we find that the alteration information is nearly in accordance with the known alteration areas. Besides, five important alteration abnormity districts have been plotted out.

关 键 词：支持向量机 主成分分析 矿化蚀变 ETM 阿哇地区 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置]