多特征匹配决策树在矿物识别中的应用  

Application of Multi-Features Matching Decision Tree to Mineral Identification

作　　者：孟佳克[1] 龚灏[1] 周仲礼[1] 芶盛[1] 

机构地区：[1]成都理工大学数学地质四川省重点实验室,四川成都610059

出　　处：《国土资源科技管理》2012年第6期93-96,共4页Scientific and Technological Management of Land and Resources

基　　金：国家863计划重点课题--地质应用系统建设与典型应用示范研究(2008AA121103)

摘　　要：利用研究区地物光谱曲线的光谱吸收指数、光谱斜率,光谱吸收深度等光谱特征参量建立多变量决策树,成功识别出白云母、方解石、绿泥石、绿帘石、蛇纹石等多种矿物。精度分析表明:建立的决策树识别结果与遥感中心的填图结果基本一致,尤其是绿帘石和蛇纹石填图结果吻合的较好,而白云母、绿泥石、方解石矿物填图结果稍有差异。This paper establishes a multi-variable decision tree by using spectral absorption indexes, spectral slopes, spectral absorption depths and other spectral characteristics, and successfully identifies various minerals, including muscovite, calcite, chlorite, epidote, and serpentine. The precision analysis shows that the identification results of the decision tree established in this paper are consistent with the mapping results of China Aero Geophysical Survey ＆ Remote Sensing Center for Land and Resources, especially the mapping results of epidote and serpentine, with slight difference in the mapping results of muscovite, chlorite, and calcite.

关 键 词：光谱曲线 光谱特征 多变量决策树 矿物 

分 类 号：P627[天文地球—地质矿产勘探]