为减小离群点对点云配准精确度的影响,避免点云配准迭代计算过程中陷入局部最小值,基于鲁棒性准则函数点云配准框架提出泰勒级数准则函数鲁棒性点云配准算法.该方法分为泰勒级数准则函数的提出和配准初始值的确定2个方面.泰勒级数准则函数中,考虑各准则函数限制离群点影响来提高配准精确度的内因,对权值递减速率较合理的Cauchy准则函数进行泰勒级数展开,构造泰勒级数准则函数解决离群值问题;配准初始值的确定中,通过计算待匹配点云数据集的重心,根据重心信息确定平移向量,解决局部最小值问题.数值实验结果表明,泰勒级数准则函数配准误差较最小二乘法、Huber、Tukey和Cauchy准则函数更小,在配准精度上有了较大的提高,并且误差值稳定收敛;引入插值算法对点云数据进行处理,对后续的配准精度有一定的改善.
In order to reduce the influence of outliers on point sets registration and avoid the local minimum in iteration,Taylor series criterion function based robust point sets registration algorithm is proposed based on the robustcriterion function point sets registration framework.The method includes the Taylor series criterion function andthe determination of the initial value of registration.Firstly,to improve registration accuracy in the present of outliers,Taylor series criterion function by Taylor series expansion for Cauchy criterion function is proposed to limitthe influence of outliers.Secondly,by calculating the center of gravity of the sets data,the initial translation vectoris gotten with the difference of the model point sets and the data point sets.With this,the local minimum valueissue of the iteration is addressed.Numerical experiments demonstrate the performance of Taylor series criterionfunction has been improved greatly in accuracy and stability compared with least squares minimization,Huber-criterion function,Tukey-criterion function and Cauchy-criterion function.The interpolation introduced to dealwith the point sets without the homonymy point also helps to improve the accuracy of the subsequent registration.