基于KDTree改进的Super 4PCS+ICP算法在点云配准中的应用研究北大核心CSCD

在点云配准过程中,为了提高点云的配准精度,针对ICP算法对于初始位姿的局限性,对点云数据进行Super4PCS+ICP的“先粗后精”处理。首先利用KDTree树搜索对应点,用局部区域的特征度确定特征点集,再使用Super4PCS算法实现粗配准。针对精配准提出KDTree树来加快速度,SVD求解对应点参数、常数为1的加权平均、求解误差函数等手段来实现对ICP算法的改进,并求出刚体变换后的旋转平移矩阵,提高点云配准精度。实验表明,相较于传统ICP算法,本文方法的配准精度有了显著的提升。本文研究的方法可为点云配准的深入研究提供一定的参考。

Study on the application of Super 4PCS+ICP algorithmbased on KDTree improvement in point cloud alignment

In the process of point cloud alignment,in order to improve the accuracy of point cloud alignment,the point cloud data is processed by Super 4PCS+ICP "coarse first and then fine" for the limitations of ICP algorithm for initial poses.The corresponding points are first searched using the KDTree,and the feature point set is determined using the feature degree of the local area,and then the coarse alignment is achieved using the Super 4PCS algorithm.KDTree is proposed for fine registration to speed up the process,SVD to solve for the corresponding point parameters,weighted average with constant 1,and solving for the error function to realize the improvement of ICP algorithm and to find out the rotation translation matrix after rigid body transformation to improve cloud registration accuracy.Experiments show that compared with the traditional ICP algorithm,the registration accuracy of the proposed method is significantly improved.The method studied in this paper can provide some reference for the further study on point cloud registration.