基于NDT和ICP融合的点云配准方法

A Point Cloud Registration Method Based on NDT and ICP Fusion

针对传统迭代最近点算法高精度低效率与正态分布变换算法高效率低精度的问题,提出了基于NDT与ICP融合的点云配准方法。首先通过NDT算法选择合适的网格参数将待匹配的点云向目标点云拉近以提高配准效率,完成粗配准,其次使用KD树加速的ICP算法求解变换矩阵以提高配准的计算效率。通过实验表明,本文方法匹配速度相比NDT算法和ICP算法有明显提高,且精度高于NDT算法。

Aiming at the problem of high accuracy and low efficiency of traditional ICP algorithm and high efficiency and low precision of NDT algorithm,apoint cloud registration method based on NDT and ICP fusion is proposed.Firstly,the appropriate point cloud is selected by NDT to draw the point cloud to be matched to the target point cloud to improve the registration efficiency,complete the coarse registration,and then use the KD tree accelerated ICP algorithm to solve the transformation matrix to improve the calculation efficiency of the registration.Experiments show that the matching speed of this method is significantly higher than that of NDT and ICP algorithms,and the accuracy is higher than that of NDT algorithm.