祁连山玉石沟地区极化SAR影像SVM岩性分类

SVM Lithological Classification of PolSAR Image in Yushigou Area, Qilian Mountain

祁连山玉石沟是重要多金属矿床产地之一，但野外地质工作程度较低，利用雷达遥感图像SAR的宏观优势和丰富地物信息识别区域岩性，有利于明确找矿方向。本文选取研究区的RADARSAT．2影像，经过预处理，对比分析不同极化影像的特征，建立不同岩性的解译标志，并结合色调、形状、纹理特征，运用面向对象的支持向量机分类方法，得到奥陶系阴沟群、火山岩、超基性岩等8类地物的分类结果，与野外调查的实际材料图对比验证，分类结果精度较高。

Concerning the scarce geological exploration of the Yushigou by human, located in Qilian Mountain, which is an important origin of polymetallic deposits, and by means of radar remote sensing image characterized by large volume information about the geology to identify regional lithology which is helpful to direct the mine; in this paper, the RADARSAT-2 data was selected after preprocessing, to establish the sample of different lithologies by comparison and analysis on the features of different polarization imaging. In addition, the eight kinds of lithologies were obtained, such as, the Ordovician sewer cluster, volcanic and ultramafic rocks, etc. using the Support Vector Machine （SVM） classification method combined with the characteristics of color, shape and texture; and then compared with the practical results from field survey, proving the high accuracy of the classification.