基于NDT与ICP的快速点云配准算法

点云配准是三维重建过程的关键一步。传统配准算法的速度较慢,尤其是在两个点云距离较远或点云数据较大的时候,为此本文提出了一种基于NDT和ICP的快速点云配准方法,能够有效地减少配准时间。本文算法主要分为三步：(1)采用NDT算法进行点云粗配准,调整两点云间的距离和点云姿态；(2)采用ICP算法对粗配后的点云数据进行微调,调整点云位置与姿态；(3)采用ICP算法对微调后的点云进行精确配准。实验结果表明,与传统算法相比,在点云数据量较大或者两个点云距离较远的情况下,本文算法也能够达到较快的配准速度与较高的配准精度。

Fast point cloud registration algorithm based on NDT and ICP

Point cloud registration is one of the key problems of 3D reconstruction. Since classical registration algorithms are relatively slow, especially to handle the point clouds with far distance or the large point clouds, a novel fast point cloud registration algorithm based on NDT and ICP is proposed. The algorithm includes three steps: (1) NDT algorithm is adopted to roughly register the point clouds and adjust the distance and attitude. (2) ICP algorithm is adopted to fine-tune the point clouds after the rough registration to adjust the position and attitude carefully. (3) ICP algorithm is adopted again to make precise registration based on the fine-tuned point clouds. The experiment shows that the proposed algorithm can effectively reduce the registration time cost and achieve high precision even if the point clouds have a far distance or include a large number of points.