基于高光谱粗糙集分析的矿化信息挖掘

Extracting Mineralization Information by Hyperspectral Analysis Based on Rough Set Theory

以野外勘察和光谱测量数据为数据源,以光谱曲线的16个特征和矿化蚀变程度分别为条件属性和决策属性,应用粗糙集理论和方法离散化数据构建决策系统和实现属性约简,获取识别矿化信息最佳变量组合及区间值,将其作为参量建立矿化信息识别模型,并以矿区其他数据进行了检验,结果与实地勘查资料基本吻合,表明该方法可以作为高光谱矿化信息识别模型,为成矿预测提供依据。

The wall rock alteration is important for information hydrothermal deposits prospecting, and so the effective extraction of mineralization information has aroused great interests among geologists in remote sensing geology. In this experiment, the data collected from field surveys and spectral measurements were analysed based on rough set theory. The decision table is established for mineralization information identi- fication by taking the 16 spectral characteristics as condition attribute set and taking degree of rock alteration as decision attribute set and discretization data. By reducing decision table to cut off redundant condition attribute, the rules between mineralization information and spectral curve features were extracted. At last, the rules were tested by the other data of mining area. The result is basically consistent with field surveys data. This shows that the method can be used as model Hyperspeetral mineralization information i- dentification, and can provide evidence for metallogenic prognosis.