摘要:
道路交叉口是道路网络(简称路网)的重要组成元素,获取道路交叉口对提取路网信息、影像匹配和车辆导航等有重要作用.然而,基于遥感图像对道路交叉口自动提取的研究尚不多.针对高分辨率遥感图像中道路交叉口的特点,提出一种自动识别道路交叉口的方法.在图像预处理的基础上,首先对图像进行多尺度圆形均匀区检测,使用梯度变换和形态学变换提取可能存在道路交叉口的候选区;然后对候选区进行特征提取和进一步筛选,得到候选道路交叉口的位置中心;最后提取位置中心的角度纹理信息,通过波谷检测判断其连接属性,识别出道路交叉口.结果表明,该方法能有效提取出城市地区的道路交叉口,对于较复杂地区的道路交叉口提取也有一定的效果.
Abstract:
Road intersection is one of the most important parts of road network,and extraction of road intersection information plays a significant role in such aspects as road network extraction,image registration and vehicle navigation.However,the research on extracting road intersections from remote imagery is insufficient.In view of the characteristics of road intersections in high resolution remote sensing imagery,the authors propose an approach to auto-extraction of road intersection in this paper.On the basis of image preprocessing,detection of homogeneous circular area by multi-scale structure elements was firstly used to extract alternative road intersections,which included gradient transformation and morphological transformation.Secondly,feature extraction for alternative road intersections was processed in order to further refine the result and get the central position for each choice.Finally,angular texture signature was extracted for each central position and road intersections were identified by valley finding.The experimental results show that the method presented in this paper can extract urban road intersections efficiently and has fairly good accuracy for complex urban context.
下载: