基于面向对象分类的城市土地利用变化检测

Urban Land Use Change Detection Based on Object-oriented Classification

以新疆乌鲁木齐某区域2013年的GeoEye影像和2015年的IKONOS影像为辅助数据,进行基于面向对象分类的城市土地利用变化检测实验;根据各类地物的提取规则,提取了实验区域的地物信息,并利用ENVI5.1软件进行了变化检测。结果表明,面向对象分类的变化检测方法总体精度为85.97%。为了提高该方法的精度,结合目视解译方法,面向对象方法 +目视解译的变化检测精度为87.00%,说明目视解译方法在一定程度上提高了检测精度。

We carried out an urban land use change detection experiment, which utilized the assistant data of GeoEye images and IKONOS images in the region of Urumqi City. According to the extraction rules of various types of features, we extracted the geomorphic information of the experimental area, and used ENVI5.1 software to detect the changes of land use. The experimental result showed that the overall accuracy of the object-oriented change detection method was 85.97%. In order to improve the accuracy of this method, we combined the visual interpretation method with it. The accuracy of the object-oriented method and visual interpretation method was 87.00%, which illustrated that the visual interpretation method could improve the detection accuracy.