一种高分辨率遥感影像城区道路网提取方法  

A methodology for urban roads network extraction from high resolution remote sensing imagery

作　　者：周家香[1] 周安发[1,2] 陶超[1] 高陈强[3] 李静[1] 

机构地区：[1]中南大学地球科学与信息物理学院,湖南长沙410083 [2]湖北省基础地理信息中心,湖北武汉430071 [3]重庆邮电大学信号与信息处理重庆市重点实验室,重庆400065

出　　处：《中南大学学报（自然科学版）》2013年第6期2385-2391,共7页Journal of Central South University:Science and Technology

基　　金：国家高新技术研究发展计划("863"计划)项目(2012AA121301);重庆市科委自然科学基金资助项目(CSTC;2010BB2411)

摘　　要：提出一种高分辨率影像城区道路自动提取新方法。该方法首先引用经典的Mean-Shift算法实现道路图像的初步分割,再合并灰度相似的区域,依据直方图准则选取合适的阈值进行二值化分割;然后,引入形状因子(面积、长宽比等)去除混杂在图像中与道路形状特征不相似的区域;对于仍然与道路相连的非道路区域,构造多方向形态学滤波的方法剔除,提取独立的道路区域,最后连接断裂的道路线,实现道路网的提取,并对多幅高分辨率城区影像进行试验。研究结果表明:该方法能很好地实现从复杂环境中提取道路网,特别是对直线型道路的提取精度更高。An automatic urban road extraction method for high-resolution images was proposed. Firstly, road images were initially segmented by using classic Mean-Shift algorithm, and then regions with similar gray values were merged, and appropriate thresholds for image segmentation were selected based on histogram analysis. Secondly, shape indices such as area, length-width ratio etc were introduced to remove those regions mixed in image which had different shapes with road. In order to ensure the independence of each road target candidate, a multidirectional morphological filtering algorithm was designed to separate road from the neighboring non-road objects. Finally, road network was extracted by connecting the broken road lines. The results indicate that this method can be used to extract roads under complex conditions, especially for the straight roads.

关 键 词：遥感影像 MEAN-SHIFT算法 形状特征 形态学 道路提取 直线连接 

分 类 号：P407.8[天文地球—大气科学及气象学]