基于光谱特征参数组合的高光谱数据矿物填图方法  

Hyperspectral Mineral Mapping Method Based on Spectral Characteristic Parameter Combination

作　　者：韦晶[1] 明艳芳[1] 刘福江[2] 

机构地区：[1]山东科技大学测绘科学与工程学院,山东青岛266590 [2]中国地质大学信息工程学院,湖北武汉430074

出　　处：《地球科学（中国地质大学学报）》2015年第8期1432-1440,共9页Earth Science-Journal of China University of Geosciences

基　　金：国家科技支撑计划课题项目(No.2012BAH27B04)

摘　　要：受大气环境等因素的影响,高光谱遥感矿物识别难以达到较高的精度.为解决该问题,根据光谱吸收特征参数在大气变化中能保持相对稳定的特点,提出一种基于光谱特征参数组合的高光谱矿物类型识别方法.文中计算了多种光谱特征参数,通过最佳指数因子(optimum index factor,OIF)优选特征参数组合,选定最佳特征参数组合,利用模式识别方法实现矿物识别.利用机载可见/红外成像光谱仪(airborne visible infrared imaging spectrometer,AVIRIS)高光谱数据,在美国内华达州Cuprite矿区进行了该方法的应用试验研究,并与前人矿物填图结果做了对比.结果表明:吸收波谷位置-吸收面积-吸收右肩位置(P-A-S2)光谱特征参数组合的矿物识别效果最优,整体精度达到74.68%.Influenced by the atmospheric environment and other factors, the mineral recognition with hyperspectral remote sensing is difficult to achieve a high accuracy. To improve the accuracy of the mineral identification with such technology, a hyperspectral mineral recognition method based on spectral characteristic parameter combination, which can maintain relatively stable characteristics with the atmospheric changes, is proposed in this paper. Various spectral characteristic parameters are calculated, and the optimal combination of the parameters is selected through the optimum index factor （OIF）, based on which, mineral identification is realized with pattern recognition method. Based on the above method, mineral type identifica- tion test is carried out in Cuprite mine of Nevada, with airborne visible infrared imaging spectrometer （AVIRIS） hyperspeetral data. The results are compared with the work of previous mineral mapping, it shows that the combination of the spectral characteristic parameters, P-A-S2 （P is absorption wave trough position, A is absorption area, S2 is absorption right shoulder position） can get the highest identification precision, the overall accuracy can reach 74.68%.

关 键 词：光谱吸收特征参数 遥感 矿物填图 机载可见/红外成像光谱仪数据 最佳指数因子 Cuprite矿区. 

分 类 号：P575.4[天文地球—矿物学] P237[天文地球—地质学]