国土资源遥感Issue(4)：21-26,6.DOI:10.6046/gtzyyg.2015.04.04
基于高分辨率影像的道路中心线提取技术研究
A study of road centerline extraction from high resolution image
摘要
Abstract
In this paper, the strategy to extract accurate road centerlines from acquired road stripe image was explored. The workflow is as follows: Firstly, road candidate points are obtained based on probabilistic boosting tree algorithm, and smooth and integrated road stripes are immediately acquaried by morphology. Secondly, thinning algorithm is introduced to automatically detect road centerlines; nevertheless, the output contained spurs and local curvature of centerlines change much. After that, geodesic distance theory is used to remove spurs. Thirdly, initial results are refined on the basis of Dijkstra algorithm. Lastly, the ultimate road centerlines are obtained according to direction consistency and road continuity. The authors performed an experiment on a high resolution aerial image. The result is satisfactory and shows that the strategy proposed in this paper is an effective method.关键词
道路中心线提取/测地距离/Dijkstra最短路径算法/概率增进树Key words
road centerline extraction/geodesic distance/Dijkstra algorithm/probabilistic boosting tree分类
信息技术与安全科学引用本文复制引用
周绍光,向晶,邱伟,孙金彦,凡莉..基于高分辨率影像的道路中心线提取技术研究[J].国土资源遥感,2015,(4):21-26,6.基金项目
国家自然科学基金项目“高分辨率遥感影像中城市道路网的提取方法研究”(编号:41271420/D010702)资助。 （编号:41271420/D010702）
