一种基于对偶四元素描述的线面特征约束的点云配准方法  

Point Cloud Registration Method Based on Dual Quaternion Description of Line-Planar Feature Constraints

作　　者：李绕波 袁希平 甘淑[1,2] 毕瑞 高莎 郭燕 Li Raobo;Yuan Xiping;Gan Shu;Bi Rui;Gao Sha;Guo Yan(Faculty of Land Resources and Engineering,Kunming University of Science and Technology,Kunming,Yunnan 650093,China;Application Engineering Research Center of Spatial Information Surveying and Mapping Technology in Plateau and Mountainous Areas Set by Universities in Yunnan Province,Kunming,Yunnan 650093,China;College of Geosciences and Engineering,West Yunnan University of Applied Sciences Dali,Yunnan 671009,China)

机构地区：[1]昆明理工大学国土资源工程学院,云南昆明650093 [2]云南省高校高原山区空间信息测绘技术应用工程研究中心,云南昆明650093 [3]滇西应用技术大学地球科学与工程学院,云南大理671009

出　　处：《光学学报》2022年第2期159-167,共9页Acta Optica Sinica

基　　金：国家自然科学基金(41861054)。

摘　　要：针对地面三维激光扫描仪采集到的相邻测站点云存在空间特征关联,以及城市建筑群中含有大量的线面特征,提出一种基于对偶四元素描述的线面特征约束的点云配准方法,该方法在利用对偶四元素统一描述空间变换参数的同时考虑到尺度因子。将依据直线和平面存在的几何关系,以及线面相交所产生的交点与夹角等作为配准约束条件,构建出空间相似变换目标函数,利用最小二乘准构建出平差模型以计算空间相似变换相关参数。为了避免不恰当的初始值所导致的迭代不收敛问题,将Levenberg-Marquardt法应用于平差模型的解算。最后对该方法的正确性与可行性进行实验分析。结果表明,相较于仅考虑线特征或面特征约束的点云配准方法,所提方法的配准精度更高,并且采用Levenberg-Marquardt法解算平差模型能在任意给定的初始值下正确收敛。In view of the spatial feature correlation of adjacent station clouds collected by ground three-dimensional laser scanners, and a large number of line-planar features in urban buildings, a point cloud registration method based on line-planar feature constraints described by dual quaternion is proposed. This method not only uses dual quaternion to describe spatial transformation parameters, but also takes into account the scale factor. The objective function of spatial similarity transformation is constructed according to the geometric relationship between line-planar, as well as the intersection point and angle caused by the intersection of line-planar as registration constraints. The adjustment model is constructed by using the least square criterion to calculate the relevant parameters of spatial similarity transformation. In order to avoid the problem of iterative non-convergence caused by inappropriate initial values, the Levenberg-Marquardt method is applied to the solution of the adjustment model. Finally, the correctness and feasibility of the method are analyzed by experiments. The results show that the registration accuracy of the proposed method is higher than that of the point cloud registration method considering only linear of planar feature constraints, and the adjustment model solved by Levenberg-Marquardt method can converge correctly under any given initial value.

关 键 词：激光光学 点云配准 线面特征约束 对偶四元素 平差处理 迭代收敛 

分 类 号：P237[天文地球—摄影测量与遥感]