基于光谱指数的遥感影像岩性分类

由于传统的岩性分类方法受岩石辐射干扰因素大, 存在"同物异谱"以及"同谱异物"现象, 岩性分类精度低, 所以在深入分析岩石矿物光谱特征基础上, 以西昆仑成矿带地区的二长花岗岩、石英正长岩以及正长岩为研究对象, 基于这3种岩性的实测光谱数据以及先进星载热发射和反射辐射仪(advanced spaceborne theemal emission and reflection radiometer, ASTER)影像数据的波段设置特征, 建立了RI和SI两种光谱指数.利用所建立的RI以及SI光谱指数对ASTER遥感数据进行岩性分类.结果显示, RI和SI两种光谱指数法在提取二长花岗岩时精度达到70%以上, 石英正长岩精度为80%左右, 与最大似然法得到的分类结果相比, 这两种岩性的分类精度明显提高了. 

Lithologic Classification from Remote Sensing Images Based on Spectral Index

In order to solve the problem of interference factors of the rock radiated in traditional lithologic classification with the phenomenon of the different spectra for the same substance and the same spectrum for different substances, and the low lithologic classification accuracy, A study based on the in-depth analysis of spectrum characteristics of mineral rock, taking monzonitic granite, quartz-syenite and syenite of West Kunlun metallogentic belt as the research object is presented in the paper. We set up two spectral index RI and SI on the basis of measured spectral data of three kinds of lithology and the characteristics of ASTER imaging data. Two spectral index methods are used to classify ASTER remote sensing data, and the results show that the accuracy in monzonitic granite extraction with RI and SI spectral index method is above 70%, about 80% in quartz-syenite extraction. The precision of the two kinds of lithology has been significantly increased compared with the classification results of maximum likelihood method.