摘要
由于航空图像的拍摄高度和拍摄环境,各种细节信息比较明显,道路受房屋、树木、车辆以及各种阴影的影响很大,根据城市道路这一特性,提出基于聚类分割的道路提取方法。先对原始航空图像做增强处理,突出图像各部分之间的差异,用K-means聚类算法对增强处理后的图像进行初步分割;对分割后的各个子图像进行Hough直线提取,再根据图像中各要素的形态特征筛选出道路中心线,随后将提取出的各道路中心线进行连接形成道路网络。实验结果表明,该方法能有效地从道路受干扰较多的航空图像中提取出城市道路网。
Because of the shooting angle and environment of aerial images, there are so many details. And the influence of buildings, tree shadows, cars and some others is great. According to such characters, a method of road extraction based on clustering segmentation was proposed. The original aerial image was enhanced to highlight the image edge information, and the enhanced image initial segmentation was made by the k-means clustering segmentation method. Hough transform was applied to every sub-image, and then the road centerline was obtained by filtering the lines extracted by Hough transform, according to the morphological characteristics of image elements. The road network was composed of road centerlines of every sub-image by connecting them together. Experimental results show that the method can effectively extract the urban road network from the aerial images.
出处
《系统仿真学报》
CAS
CSCD
北大核心
2014年第9期2198-2202,共5页
Journal of System Simulation