中高分辨率遥感数据铁染信息提取及找矿预测

Extraction of Iron Stained Information and Ore-prospecting Prediction of Medium-high Resolution Remote Sensing Data

苍山地区铁矿资源丰富,但地表覆盖较重,在区域范围内常规方法提取蚀变信息效果差。针对此问题,结合使用ASTER与GF-1卫星影像数据,从分析典型矿物光谱吸收特性出发,综合影像色彩、波段特征等多种信息采用了"主成分分析+波段比值+假彩色合成"和"主成分分析+密度分割"2种蚀变信息提取模型。方法主要运用了主成分分析思想去除冗余数据,前者辅助波段比值及彩色合成技术突出铁染信息,后者以选取最佳阈值的方式对高分辨率数据进行密度分割,优化了提取效果。2种方法结合有效提取了铁染蚀变信息,综合提取结果与区域控矿断裂特征,圈定了4处成矿靶区。对研究区进行实地采样验证,Kappa系数分别为0.826和0.87,精度较高。该方法的实现为同类地区的找矿工作提供了有利依据。

Though with rich mineral,the surface mulch of Cangshan area was much heavier which leads to the poor result in extracting alteration information by regular method within the area.To tackle with this problem,this paper,by combination use of ASTER and GF1remote sensing image,beginning with absorptive behavior analysis of typical mineralized spectral and integrating with various information,such as image color as well as band characteristic,etc.,adopted two extraction patterns for alteration information,including“principal component analysis+band ratio+false color composition”and“principal component analysis+density segmentation”.The above two patterns predominantly applied with the thought of principal component analysis,aiming to remove the redundant data,of which the prior was to highlight the iron stained information as a supplementary of band ratio and color composition technique,while the latter was to realize the density slicing of high resolution data through selecting the optimal threshold,thus to optimize the extraction efficiency.Finally,four minerogenic prospects were delineated by combination of regional ore controlling fault features and extraction result of iron stained information which was effectively obtained through binding of the above two patterns.The field sampling in the research area proved that Kappa coefficient was0.826and0.87respectively,showing higher accuracy.The implementation of these two methods provided beneficial basis for mineral prospecting work in the similar area.