一种基于全等三角形的点云自动配准方法  

An Automatic Registration Method of Point Clouds Based on Congruent Triangles

作　　者：代许松 花向红[1,2] 任志忠 陶武勇 赵不钒 李琪琪 DAI Xusong;HUA Xianghong;REN Zhizhong;TAO Wuyong;ZHAO Bufan;LI Qiqi(School of Geodesy and Geomatics,Wuhan University,Wuhan 430079,China;Disaster Monitoring and Prevention Research Center,Wuhan University,Wuhan 430079,China;Guangzhou Changdi Spatial Information Technology Co.,Ltd.,Guangzhou 510663,China;School of Information Engineering,Nanchang University,Nanchang 330031,China)

机构地区：[1]武汉大学测绘学院,湖北武汉430079 [2]武汉大学灾害监测与防治研究中心,湖北武汉430079 [3]广州长地空间信息技术有限公司,广东广州510663 [4]南昌大学信息工程学院,江西南昌330031

出　　处：《测绘地理信息》2023年第5期55-59,共5页Journal of Geomatics

基　　金：国家自然科学基金(41674005,41871373);广东省级促进经济高质量发展专项(GDNRC[2020]050)。

摘　　要：点云配准是点云数据处理中比较关键的步骤,直接影响处理结果。经典的迭代最近点(iterative closest point,ICP)算法需要目标点云与源点云之间有良好的初始姿态,否则会遇到局部最优等问题。因此,提出了一种基于全等三角形的点云自动配准方法。该方法通过边长相等和面积相等来构造全等三角形,找到源点云与目标点云的对应点,建立源点云与目标点云的对应关系和转换参数最优估计,完成粗配准,再结合ICP算法进行精配准,实现点云的自动配准。结果表明,与四点一致集(4-points congruent sets,4PCS)配准算法和ICP算法的结合算法相比,所提算法能有效改善ICP算法对初值依赖的问题,并且配准精度有一定提升。Point cloud registration is a key step in point cloud data processing,which directly affects the quality of the processing results.The classic iterative closest point(ICP)algorithm requires a good initial pose between the target point cloud and the source point cloud,otherwise some problems such as local optimization will occur.Therefore,we propose an automatic registration method of point clouds based on congruent triangles.First,congruent triangles are constructed by equal side lengths and equal areas,then the corresponding points between the target point cloud and the source point cloud are determined.The correspondence between them and the optimal estimation of the transformation parameters are established to complete coarse registration.Then,ICP algorithm is used for the fine registration.The results show that the dependency of ICP algorithm on initial values can be effectively reduced and the registration accuracy is improved by the proposed algorithm,compared with the algorithm combining 4-points congruent sets(4PCS)registration algorithm and ICP algorithm.

关 键 词：激光雷达(light detection and ranging LiDAR)点云 全等三角形 点云粗配准 迭代最近点(iterative closest point ICP)算法 点云精配准 

分 类 号：P237[天文地球—摄影测量与遥感]