采用代表点插值的道路网提取方法

Road Network Extraction Method Using Representative Points Interpolation

针对浮动车轨迹数据包含较多异常数据导致道路网提取效果不好的问题,提出一种渐近方式的代表点插值算法.首先,采用DBSCAN聚类算法对转弯点聚类求出路口位置,计算经过两个路口间的轨迹,合并轨迹.然后,采用改进的代表点算法提取代表点.最后,利用最短路算法在由代表点建立的Delaunay三角网上插值,从而得到道路网.实验结果表明:该方法能有效地从带有较多异常的轨迹数据中取出由复杂网络构建的道路网,具有较好的实用性.

Aiming at the problem of the floating car trajectory data which contains many abnormal data leading to the bad road network extracting,the paper proposes a way of asymptotic representative point interpolation algorithm.Firstly,the method uses the DBSCAN clustering algorithm to find the intersections position,computes the trajectories between two intersections,and then merge the trajectories.Then an improved representative point algorithm is used for the extraction of representative points.Finally,using the shortest path algorithm to interpolate in Delaunay triangulation net which is established according to the representative points,road network is obtained.The experimental results show that the road network constructed by complex network can be extracted with the trajectory data which contains abnormal data,and has good practicability.