提取城市道路边线的点云法向量聚类法

Extraction City Road Boundary Method Based on Point Cloud Normal Vector Clustering

考虑城市道路边缘的空间姿态特征,建立了一种自动提取道路边线点云并绘制道路边界线的方法.建立空间格网点云存储结构,利用网格内整体点云的法向量特征对其进行聚类和分割,利用改进K均值算法提取道路边线点云,采用中值法绘制道路边界线.基于本文方法,分别选择实验区直线段和曲线段道路扫描数据进行分析.结果表明,本文方法不依靠任何辅助信息,便可以提取直线和曲线道路边线点云以及绘制道路边界线,在路面平坦的工况下有较好的提取效果.

According to the spatial posture characteristics of the urban road boundary, the method of automatically extracting road boundary point cloud and drawing road boundary lines was set up. Firstly,the spatial storage structure of the space grid point cloud was established,and used the normal vector feature of the whole point cloud in the grid to cluster and split them,and then the improved K-means algorithm was used to extract the road boundary point cloud. Finally,the road boundary line was drawn by the median method. Based on the method,the data of the straight section and the road segment scanning data of the experimental area are selected for analysis. The results show that the proposed method can extract the point cloud of straight line and curve line road boundary and draw the road boundary line without any auxiliary information,it has better extraction effect under the flat condition of the road.