面向对象分类方法在铁尾矿堆快速提取中的应用研究

摘要

以黄石市大冶铁矿区为例,利用面向对象分类方法进行铁尾矿堆信息快速提取试验研究.首先,根据WorldView-2影像特点,充分利用其丰富的光谱特征及精确的空间形状特征进行图像分割,突出影像对象边缘、重现地物实际存在情况；其次,分析影像对象的光谱、形状、纹理、拓扑关系等特征信息,建立分类规则进行分类,提取出尾矿堆信息.为了进一步提高分类精度,可以利用eCognition软件RS/GIS数据集成功能,在面向对象分类结果上进行目视解译.试验证明,面向对象分类方法适用于提取矿区尾矿堆信息,是高分辨率遥感影像自动分类的理想选择.%Taking Daye iron deposit in Huangshi as an example,it extracted the information of the iron tailing pile with object-oriented classification method. First of all, the WorldView-II image was segmented by its rich spectral feature and precise special shape feature to enhance the edge of objects and the relative location of the objects; secondly,analyzing the spectrum, shape,texture and topological relation of the objects,to establish the classification rule,and then extracted the gangue information. In order to improve the accuracy of classification,it could use the RS/GIS data integration function of eCognition.to make visual interpretation based on results of object-oriented classification. The experiment had proved that the object-oriented classification method is appropriate for extracting the information of iron tailing pile and it is the best choice of automatic classification of high-resolution remote sensing image.