基于支持向量机的矿化蚀变信息提取研究——以青海黄南州吉地地区为例  

Extracting Alteration Information Based on SVM——Taking Jidi Area as an Example

作　　者：薛云[1] 戴塔根[1] 邹林 夏浩东[3] 刘江龙[1] 

机构地区：[1]中南大学地学院,长沙410083 [2]有色地质调查中心,北京100814 [3]国土资源部实物地质资料中心,北京101149

出　　处：《地质调查与研究》2007年第4期315-320,共6页Geological Survey and Research

基　　金：国家国土资源大调查项目:矿产资源遥感综合信息提取技术与找矿应用研究(1212010660601)

摘　　要：支持向量机(SV)M引入遥感图像处理领域并逐步得到推广,但在遥感地质应用中则刚刚起步。本研究运用SVM提取矿化蚀变信息,选择青海黄南州吉地地区作为典型研究区,首先选取训练样本,然后求解最优超平面(即找出支持向量),进而确定决策函数,最后泛化推广识别其它待识别的样本。矿区叠加研究和野外实地验证表明SVM提取矿化蚀变信息克服了传统的统计方法需要大量样本的缺陷,保证了矿化信息提取的精度。最后,指出了三个重点异常区。Support Vector Machine （ SVM ） is a new prospecting technique in remote sensing, and it＇s application has just begon in remote sensing geology. Alteration information was extracted by using SVM method recently in the author＇s work area, Jidi, Huangnan city, Qinghai Province. Firstly, training sample was selected. Secondly, the most optimal hyperplane and decision-making function was found. In the end, alteration information was extracted. Through the field investigation, we find that the alteration information is nearly in accordance with the known alteration areas after comparing the result with the data of the known alteration areas and the mineral geology mapping of the field. Besides, 3 important alteration abnormity areas have been plotted out for the next prospecting.

关 键 词：支持向量机 矿化蚀变 光谱特征 ETM 吉地地区 

分 类 号：P627[天文地球—地质矿产勘探]