一种改进的稀疏迭代最近点算法

摘要

稀疏迭代最近点算法是针对合有噪声点的点云配准提出的,但它却存在对目标点云中的离群点敏感、运行效率低等问题.针对这些问题,基于邻域信息的对应点对寻找方法提出了一种改进的稀疏迭代最近点算法.改进的稀疏迭代最近点算法首先使用改进的基于PCA的点云初始配准调整两片点云的位置,而后使用基于邻域信息的对应点对寻找方法为精配准寻找对应点对,针对对应点对,使用乘法器的交替方向法(ADMM)求得最优的变换矩阵.实验表明,对含离群点的斯坦福兔子、盆栽等点云来说,改进后的算法能够处理目标点云含有离群点的情况,并且算法的配准速度平均提高了30％.%The sparse iterative closest point algorithm for point cloud with noise points is sensitive to the outliers contained in the target point cloud,and is inefficient.To solve the problems,we find the corresponding point-pairs based on neighborhood information to improve the sparse iterative closest point algorithm.The improved sparse iterative closest point algorithm firstly uses the improved registration based on the PCA to adjust the position of the two point clouds,and then finds the corresponding point-pairs based on neighborhood information.Finally we use the alternating direction method of multipliers (ADMM) to get the optimal transformational matrix for corresponding point-pairs.Experiments on Stanford rabbit and potted model show that the improved algorithm can handle the outliers contained in the target point cloud,and the algorithm speed can be increased by 30％.