基于SAC-IA和改进ICP算法的点云配准技术

摘要

点云配准是真实三维世界物体或场景模型重建的关键问题之一.针对传统的ICP算法收敛速度慢,且在两点云集初始位置较大时易陷入局部最优解的问题,提出一种改进的点云配准算法.该算法首先利用基于特征点的采样一致性初始配准算法(SAC-IA)实现两点云的初始变换,使两点云集有相对较好的初始位姿,然后在传统ICP算法基础上使用k-d树(k-dimensional tree)加速对应点对的查找速度,并利用方向向量阈值去除错误点对.实验证明该算法具有相对较好的配准精度和收敛速度.%Point cloud registration is one of the key issues in real three-dimensional world objects or scene model reconstruction.The convergence rate of ICP algorithm is slow.When the two positions are large,the local optimal solution is caught.In response to this problem,an improved ICP algorithm is proposed.The algorithm first uses SAC-IA to realize the initial transformation of the two-point cloud,so that the two points can be in a relatively good initial position.And then the k-dimensional tree and the direction vector threshold are used on the basis of the traditional ICP algorithm.The k-d tree is used to speed up the search speed of the corresponding point pairs.The direction vector threshold is used to remove the error corresponding point.Experiments show that the algorithm has a relatively good registration accuracy and convergence speed.