基于连续点云数据的既有铁路轨面信息快速提取算法设计

Algorithm design for fast extraction of rail-surface information for existing railway based on continuous point cloud data

三维激光扫描获取的既有铁路点云数据具有海量性、离散性等特点,难以从点云数据中快速提取线路参数。为此,结合现场实测数据,提出一种基于连续点云数据的既有铁路轨面信息快速提取算法。通过k-d树实现点云的快速搜索、查询和储存,利用主成分分析法和移动激光点聚类法提取的接触线分割构建出铁路缓冲区,进而通过平面格网法的粗提和多种约束条件下的精提实现了轨面点提取。对既有线路现场试验结果表明,轨面点提取的完整度c和准确度p均在93%以上,该方法能较好地实现钢轨轨面点云的快速提取。

The existing railway point cloud data obtained by 3D laser scanning has the characteristics of massiveness and discreteness,and it is difficult to quickly extract circuitparameters from the point cloud data.For this reason,combined with the field measured data,a rapid extraction algorithm for existing railway rail surface information based on continuous point cloud datawasproposed.Point clouds fast search,inquiry and storage were realized through k-d tree,and the railway buffer zonewas split and constructed by using the contact line,which extracted by principal component analysis and moving laser point clustering method.Then the rail surface point extraction was realized by rough extraction of plane grid method and fine extraction under various constraints.The field test results of existing lines show that the integrity c and accuracy p of rail surface point extraction are above 93%,and this method can better achieve rapid extraction of rail surface point cloud.