Abstract: There has been a lot of methods that withdraw the road from remotely sensed image，such as image segmentation，road characteristic identifying basing on knowledge and morphology.But there are still many problems need to be resolved.This text carried out an algorithm.The first step involves road image two values，removing noise data，edge examination and edge conjunction using morphology.Through experiments using the ETM+data，this algorithm have a better anti-noise ability and higher accuracy，comparing with the traditional algorithm that edge detecting and edge conjunction.This algorithm has a certainly realistic meaning.

Key words: morphology, edge detection, edge conjunction, road detection

摘要： 利用遥感图像进行道路提取已经有了一些研究，如图像分割，基于知识的道路特征识别和数学形态学等，但尚有许多问题需要解决，设计了一种基于数学形态学的遥感图像道路提取算法。该算法首先将遥感图像二值化，然后进行噪声滤除、形态学边缘检测与边缘连接。通过采用ETM+遥感图像进行实验表明，该算法与传统的边缘检测与边缘闭合的算法相比，具有更好的抗噪能力，且精度较高，具有一定的现实意义。

关键词: 数学形态学, 边缘检测, 边缘连接, 道路提取