Multi-stage road three-dimensional laser point cloud accurate registration method

The invention discloses a multi-stage road three-dimensional laser point cloud accurate registration method. Point cloud data is segmented according to a certain distance to obtain multiple segments of data, then the multiple segments of data are registered respectively, deviation correction perpendicular to the direction of a driving lane is conducted firstly, then deviation correction in the direction of the driving lane is conducted, finally elevation correction is conducted, and finally registered data are obtained. According to the invention, segmentation processing is carried out on thelarge-scale scene point cloud data, and independent ground objects and road markings are used as registration elements, so that the registration process is optimized, and the problem of low speed of apoint-to-point registration mode of massive point cloud data is solved. 本发明公开了一种多期道路三维激光点云精确配准方法,它先将点云数据分别按一定的距离进行分段，得到多段数据，接下来对多段数据分别配准，首先进行垂直于行驶车道方向的偏移纠正，其次进行行驶车道方向的偏移纠正，最后进行高程纠正，得到最终配准后的数据。本发明对大场景点云数据进行分段处理，同时采用独立地物和道路标线作为