基于PCL的点云数据粗配准算法研究

PCL-Based Coarse Registration Algorithm for Point Cloud Data

摘要: 传统ICP算法精度受点云初始位姿影响较大，收敛速度慢，不能满足精细化点云建模的要求。基于此问题，通过基于快速点特征直方图的采样一致性配准方法进行粗配准。首先将两帧待配准点云进行体素滤波，其次进行表面法向量估计并计算关键点的PFPH特征，然后运用采样一致性算法得到最优变换，最后在此基础上再进行ICP配准。实验表明，该方法能有效改善配准精度和收敛速率。

Abstract: The accuracy of the traditional ICP algorithm is greatly affected by the initial pose of the point cloud, and the convergence speed is slow, which cannot meet the requirements of refined point cloud modeling. Based on this problem, the coarse registration is performed by the sampling consistency registration method based on the fast point feature histogram. Firstly, two pieces of the experimental point cloud to be registered are subjected to voxel filtering, followed by surface normal vector estimation and calculation of the fast point feature histogram of the key points, and then the sampling consistency algorithm is used to obtain the optimal transformation. Finally, the ICP matching is performed on the basis of this quasi. Experiments show that this method can effec-tively improve the registration accuracy and convergence rate.