基于多角度点云数据的快速配准方法

A Rapid Registration Method Based on Multi-Angle Point Cloud Data

针对点云数据三维重建过程中多个角度的点云数据匹配运算时间过长的问题,提出了一种基于计算机集群的多角度点云数据快速配准方法,通过对传统点云配准方法的并行化设计以达到快速配准的目的.该方法不仅能够获取到正确的点云配准结果,而且面对的运算量越大越能体现出其优越性.据此,以单个角度约50万个顶点的点云数据进行了配准实验,实验表明,随着需要进行匹配的点云数据从2组增加到6组,在传统的配准方法运算时间大幅度提升的情况下,此配准方法的加速比从0.98增加到3.91,有效缩短了数据配准运算所需的时间.

In order to solve the problem of huge time cost in the registration process of multi-angle point cloud data, the rapid calculation method is represent in this paper based on computer cluster, which can parallel the traditional registration method of point cloud data. The method can not only acquire the right calculation results, but also show its advantage while the computation burden gets heavier. Experiments were done based on point cloud data with 500000 points in one-single angle. Experiments results show that the rapid method ~rom this paper can provide a speedup ratio from 0. 98 to 3. 91 while the point cloud data angle increase from 2 to 6. At the meantime, the traditional registration method lead to huge computation time cost. The experiments results prove that the method represent in this paper can reduce the registration time of point cloud data efficiently.