Abstract: According to the characteristics of urban road in high-resolution remote sensing image, this paper presents an approach to extract the urban road from high-resolution remote sensing images based on Hough transform and path morphology. The method uses Hough transform to detect the direction and the length of main roads, and to construct the adjacency graph, according to each direction adjacency graph determine the path. These paths can be regarded as narrow, the elongation but not completely straight structure elements. Experimental results show that it not only can detect straight road information easily, but also can detect the wide crooked road information and the curvature of smaller slender road information.

Key words: Hough transform, path morphology, remote sensing image, high-resolution, urban road extraction, adjacency graph

摘要： 针对高分辨率遥感影像上城区道路的特点，提出一种基于Hough变换和路径形态学的高分辨率遥感影像城区道路提取方法。用Hough变换检测主要道路的方向及长度，建立邻接图，根据邻接图确定路径的方向和长度，将这些确定的路径看作狭长但并非完全笔直的结构元素。实验结果表明，该方法不仅能够检测出直线道路信息，还能检测出宽的弯曲的道路及细长的曲率比较小的道路信息。

关键词: Hough变换, 路径形态学, 遥感影像, 高分辨率, 城区道路提取, 邻接图

CLC Number: 

• TP391.41