Abstract: Road extraction from remote sensing image is always the hot and difficult problem for information extraction，the paper extracts road combining remote sensing classification and mathematic morphology.Remote sensing classification can divide artificial object and nature object，it will avoid the nature object effect during the road extraction.The erosion and dilation operator of mathematic morphology is very sensitive to the object edge.The result shows that the method can extract road information effectively.

摘要： 利用遥感图像获得完整道路信息网一直是信息提取的热点和难点，研究了结合遥感分类和数学形态学提取道路，以期获得满意的结果。遥感分类能有效区分人工地物与自然地物，因此能避免自然地物对提取的影响；数学形态学中的腐蚀和膨胀算子对地物边缘很敏感，一直是边缘信息提取的主要方法。实验表明该思路能快速有效提取道路信息。