机载LiDAR点云提取城市道路网的半自动方法  

A semi-automatic algorithm of extracting urban road networks from airborne LiDAR point clouds

作　　者：李峰[1] 崔希民[2] 刘小阳[1] 卫爱霞 

机构地区：[1]防灾科技学院,河北三河065201 [2]中国矿业大学(北京)地球科学与测绘工程学院,北京100083

出　　处：《测绘科学》2015年第2期88-92,共5页Science of Surveying and Mapping

基　　金：中央高校基本科研业务费专项资金青年教师资助计划项目(ZY20140211);国家自然科学基金项目(41071328)

摘　　要：针对从卫星和航空正射影像中手工提取道路的方法速度慢的问题,该文基于机载LiDAR点云的高程和反射强度信息,手工选取一定量的种子点,采用区域生长法分割出初始道路区域;在填充空洞和平滑预处理后,利用细化算法提取出道路中线;根据长度阈值剔除毛刺短分支,使用自动算法和手工结合的方法消除多余通路和环形通路;并跟踪连接成矢量道路线,根据抽稀算法移除大量的小弯曲拐点,最后平滑矢量道路线成道路网。在测试复杂场景下的城市LiDAR点云后,发现提取的道路网的完整率和正确率较高。Based on height and intensity of airborne LiDAR point clouds,some seed points were selected by manual and initial road sectors were segmented by region growing method in the paper.After enforcing filling holes and smoothing approaches,a thinning algorithm was used to extract centerlines of roads.The noise short branches were removed according to a length threshold.Combing an automatic algorithm with manual way,redundant pathways were wiped off.Then,this algorithm tracked and connected pixels into vector lines of roads.Plentiful inflection points having small curve were identified and deleted using rarefying method.Finally,vector road network was retrieved by smoothing method.Experimental result by the LiDAR point clouds from a complex urban area showed that the completeness and correctness of extracted roads by the method could be good.

关 键 词：机载LIDAR 点云 半自动 道路网 提取 

分 类 号：P234.1[天文地球—摄影测量与遥感]