Abstract: In order to extract roads automatically from high resolution remote sensing images with effect, a method based on the topology of lineaments to recognize lineaments on road is proposed. Line segments as well as gray values on their left and right sides are extracted from remote sensing images; line segments on roads are clustered as road lineaments using the recognition model proposed in this paper; verifications are done using prior knowledge. The method mainly uses struct information of lineaments, and it is robust to be applied to road extraction from remote sensing images with large scales and complex scenes. The method has been applied to the production and update of road data for GPS navigation from high resolution remote sensing images.

Key words: high resolution remote sensing, linear feature, road extraction, road recognition, topology

摘要： 为有效从高分辨率遥感影像中自动提取道路，提出利用线性要素间的拓扑关系识别道路线性要素的方法。提取影像中的线性要素并获取其邻域的光谱属性形成有向直线段；利用提出的道路线性要素识别模型，将满足条件的有向直线段聚类生成道路要素集；利用先验知识进一步验证。该方法主要使用道路线性要素的结构信息，适用于大范围、场景复杂的遥感影像，具有较高鲁棒性，目前该方法已应用于基于高分辨率遥感影像的GPS导航道路数据的生产和更新。

关键词: 高分辨率遥感影像, 线性特征, 道路提取, 道路识别, 拓扑