混合稀疏迭代最近点配准  

Sparse mixture iterative closest point registration

作　　者：刘跃生 陈新度[1] 吴磊[1] 黄运保[1] 李海艳[1] LIU Yue-sheng;CHEN Xin-du;WU Lei;HUANG Yun-bao;LI Hai yan(College of Electromechanical Engineering,Guang Dong University of Technology.,Guangzhou 510006,Guangdong)

机构地区：[1]广东工业大学机电工程学院,广东广州510006

出　　处：《光学精密工程》2021年第9期2255-2267,共13页Optics and Precision Engineering

基　　金：广州市科技计划项目(No.201902010054);国家自然科学基金(No.51775116,No.51975125)。

摘　　要：为了避免离群值的影响,提出了混合稀疏迭代最近点(SM-ICP)方法,以实现点云精确配准。本文对稀疏表示、正则化求解和点云配准方法进行了研究。首先,利用混合正则项表示配准残差,构建混合稀疏配准函数。然后,结合交替乘子法(ADMM)构建了所提出函数的双循环优化框架。其中,混合正则项的平衡权重θ可由Sigmoid函数求解;此外,还给出了ADMM优化框架内循环中对应损失函数的标量形式。最后,推导了该标量化损失函数在点云配准中的软阈值表达式。实验结果表明,所提出SM-ICP算法的配准精度优于所对比的算法。特别的,在重叠率约为50%的斯坦福兔子配准实验中,SM-ICP算法的截断配准误差为2.04×10^(-4),较Robust Trimmed-ICP(Tr-ICP)算法和ICP算法的配准误差减小了一个量级,且较稀疏ICP(S-ICP)算法减小了约三倍;在其它对象、场景类型的点云配准实验中,SM-ICP算法的配准精度同样较其它对比算法更优;在具有不同层级随机噪声点云的配准实验中,SM-ICP的截断配准误差为4.90×10^(-6)~1.33×10^(-4),同样较其它对比算法减小了一个量级或几倍;在发动机叶片配准实验中,本文方法成功实现了点云精确配准,而其它对比算法的配准结果中存在不同程度的点云错位情况。所提出的点云配准方法具有精确、鲁棒性和泛化性等优势。The sparse mixture iterative closest point(SM-ICP)method is proposed for achieving accurate alignment of point-sets,while avoiding the influence of outliers.This study investigates sparse representa⁃tion,non-convex optimization,and point-sets registration.First,the registered residuals are represented by mixed regularization to establish a sparse mixture formula.The alternating direction of multiplier meth⁃od(ADMM)is then integrated to solve the proposed formula using a nested framework.Among the vari⁃ables,the balance weightθfor mixed regularization can be calculated using a sigmoid function.The scalar version is also provided to represent the corresponding loss of function in the inner loop of ADMM.Final⁃ly,the soft threshold formula for the scalar version can be deduced in point-set registration.Experimental results indicate that the registration accuracy of the proposed SM-ICP method is better than the that of es⁃tablished algorithms investigated for comparison.This improved accuracy is especially striking in the registration experiment of the Stanford bunny dataset.With 50%overlap rate,the trimmed registration error of SM-ICP was 2.04×10^(-4).Compared with other methods,our trimmed error was one order of magnitude lower than those of the robust Trimmed-ICP(robust Tr-ICP)and ICP algorithms.Moreover,it was ap⁃proximately three times lower than the error obtained using the sparse ICP(S-ICP)algorithm.In the reg⁃istration experiments for both other objects and for scene data,the registration accuracy of the SM-ICP method also performed better than comparable algorithms.In the registration experiment of point-sets with different levels of random noise,the trimmed registration error of SM-ICP was 4.90×10^(-6)~1.33×10^(-4).This was several times to one order of magnitude lower than those of other algorithms.In the registration experiment for the engine blade,our method successfully achieved accurate registration of point-sets,but the results produced by comparable algorithms displayed different deg

关 键 词：点云配准 迭代最近点算法 混合稀疏表示 正则化 交替乘子法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TN249[自动化与计算机技术—计算机科学与技术]