摘要
针对迭代最近点(Iterative Closest Point,ICP)算法中由噪声带来的配准效果不佳或失败的问题,论文提出了一种有效的含噪声点云的配准算法,即概率迭代最近点算法(Probability Iterative Closest Point,PICP)。首先,建立两个点云之间的一一对应关系,以提高算法的配准精度;然后,采用高斯概率模型解决刚体变换的问题,由此实现两个含噪声点云的精确配准。实验结果表明,该算法不仅能够快速准确地实现同一物体不同角度的带有噪声点云的配准,还能准确有效地实现刚体碎块间的断裂面的完全匹配和部分匹配问题,是一种准确、快速的配准,并能有效避免噪声和外点的干扰,适用范围更加广泛。
Aiming at the failure registration of iterative closest point(ICP)algorithm brought by noise,the paper proposes a point cloud registration algorithm with noise based on Expectation Maximum(EM)estimation,which is named probability iterative closest point(PICP)algorithm.Firstly,apoint-to-point correspondence is built between two point clouds,thus the registration accuracy is improved greatly.Then,Gaussian model is introduced into ICP algorithm,the singular value decomposition(SVD)method is used to solve the problem of rigid body transformation,thus the accurate registration of two point clouds is completed.The experimental results show that PICP algorithm not only can complete point clouds registration with noise of the same object from different angles rapidly and accurately,but also can achieve complete matching and partial matching of fracture surfaces between rigid body blocks effectively.It is an accurate and fast algorithm which can effectively avoid noise and external interference.It has more extensive application scopes.
出处
《计算机与数字工程》
2017年第3期419-422,522,共5页
Computer & Digital Engineering
基金
国家自然科学基金项目(编号:61373117)资助
关键词
点云配准
迭代最近点
高斯模型
概率
噪声
point cloud registration
iterative closest point
Gaussian model
probability
noise