基于几何特征的点云配准算法

为了有效地解决不存在明确对应关系的点云配准问题，提出了一种基于点云几何特征的配准算法。首先以点云的曲率为联系特征，搜索配准点云的匹配对集合；然后利用邻域特征对各匹配对进行相似性度量，提取有效配准对，并引入刚体变换中向量几何性质剔除其错配对，生成点云初变换；最后采用ICP算法对点云初配结果进行优化，实现点云精确配准。仿真实验结果表明：该算法具有较高的配准精度，且配准时间较短，是一种可行的点云配准算法。

Registration of Point Clouds Based on Geometric Properties

Aiming at the problem of point clouds registration without prior information on transformation, a novel registration algorithm is proposed based on geometric properties of point clouds. Firstly, all the pair wise points are searched by taking the curvature of point clouds as the registration relationship. Secondly, by using Euclid distance to match the pair wise points, and by introducing the geometric properties of vectors of rigid body transformation to eliminate the mismatch points, this algorithm can obtain the effective coupling points to compute original transform matrix. Finally, by using ICP algorithm to modify the former result, the optimal registration can be achieved. Experimental results show that the proposed algorithm is robust and can register the point clouds of different scans.