一种高分辨率遥感影像快速自动道路提取方法

A Quickly Automatic Road Extraction Method for High-Resolution Remote Sensing Images

摘要: 高分辨率遥感影像为用户提供丰富的地表细节信息，而如何利用这些细节信息获取地理目标，更新地理信息数据库，是遥感信息处理研究的热点也是难点之一。本文提出了一种高分辨率遥感影像快速自动提取道路的方法，这种方法以影像的分割和线特征提取为基础，综合利用道路的面状特征和线状特征，并利用面状特征道路对象应具有足够数量的线特征信息以及道路对象应具有较小形状指数的判断原则来提取和剔除道路对象。实验结果表明，这种方法能够从影像中较好地提取乡村的常规线性道路和城区具有较多面状区域的非常规道路。

Abstract: High-resolution images provide abundant surface details for users, so how to extract geographic objects with these details, to update geographic information database has become the issue of re-mote sensing information processing. This paper presents a fast automatic road extraction method of high-resolution remote sensing images. This approach, based on the fast image segmentation and line extraction, comprehensively utilizing the planar and linear feature information of the road, uses the judgment rule of planar road objects with sufficient line information and road objects with smaller shape index to extract and remove a road object. Experimental results show that this method can well extract conventional rural linear road and unconventional urban planar road.