提出了一种基于余弦相似度的点云配准(PCR-CS)算法,该算法主要解决点云刚性配准问题,即找到点云配准的旋转矩阵R和平移矩阵T,从而实现原始点云P到目标点云Q的配准。先对两个待配准点云进行去中心化处理,再进行点云余弦相似度的研究,将两个待配准的三维点云分别投影到XY平面上,对XY平面上的点云进行栅格化处理,统计栅格上的数据点从而形成统计矩阵SP和SQ,采用差分进化算法,以两点云余弦相似度为条件,寻求最优R,从而实现点云配准,最后,利用中心点计算T。实验结果表明,与其他算法相比,该算法具有较高的配准精度,即使在点云数据伴随有噪声和数据缺失的情况下,也都能达到良好的配准效果。

A point cloud registration algorithm based on cosine similarity (PCR-CS) is proposed. This algorithm mainly solves the problem of point cloud rigid registration, which involves finding the rotation matrix R and the translation matrix T of the point cloud registration to realize registration between the original point cloud P and the target point cloud Q. In the proposed algorithm, first, the two points clouds to be registered are decentralized and the cosine similarity of the point clouds is studied. Then, the two three-dimensional point clouds to be registered are projected onto the XY plane and rasterized on the XY plane. The data points on the statistical grid form the statistical matrices S_P and S_Q. Moreover, the differential evolution algorithm is used to find the optimal R under the condition of the cosine similarity of the two points clouds to achieve point cloud registration. Finally, the center point is used to calculate T. Experiment results show that compared with other algorithms, the proposed algorithm has higher registration accuracy. In addition, even when the point cloud data are accompanied by noise or missing data, it can achieve good registration results.