复杂场景下多模态点云数据配准技术  

Multi-modal point cloud data registration technology in complex scenarios

作　　者：付超[1] 夏佳毅[1] 解琨[1] 吴大鹏[1] 付沁珵 FU Chao;XIA Jiayi;XIE Kun;WU Dapeng;FU Qincheng(Jiangsu Provincial Basic Geographic Information Center,Nanjing 210013,China;Jiangsu Coal Geological Survey Team,Nanjing 210046,China)

机构地区：[1]江苏省基础地理信息中心,江苏南京210013 [2]江苏煤炭地质物测队,江苏南京210046

出　　处：《测绘通报》2024年第6期146-150,共5页Bulletin of Surveying and Mapping

基　　金：江苏省自然资源科技项目(2021052)。

摘　　要：针对复杂环境下多模态点云数据获取难,以及对点云数据配准、三维模型构建精度的要求越来越高的情况。本文以南通大剧院实景三维建模为例,当初始点云和校准点云两组多模态融合点云位置差较大时,采用ICP算法进行点云配准易导致局部最优问题,利用所提出的基于控制点辅助约束的最近点迭代(CPA-ICP)算法通过对点云数据进行配准,并与其他3种点云配准算法的试验进行对比,可知该方法的配准精度和配准效率较高,对复杂场景下的多模态点云数据融合有较好的参考意义。Due to the difficulty in obtaining multi-modal point cloud data in complex environments,the requirements for point cloud data registration and 3D model construction accuracy are becoming increasingly high,this article takes the real-life 3D modeling of Nantong Grand Theater as an example.When the position difference between the initial point cloud and the calibration point cloud is significant,using the ICP algorithm for point cloud registration can easily lead to local optimization problems.The proposed nearest point iteration(CPA-ICP)algorithm based on control point auxiliary constraints is used to register point cloud data,and the experimental comparison of the other three point cloud registration algorithms fully demonstrates,This method has high registration accuracy and efficiency,and has good reference significance for multimodal point cloud data fusion in complex scenes.

关 键 词：复杂场景 多模态点云 联合定向匹配 CPA-ICP算法 数据融合 

分 类 号：P23[天文地球—摄影测量与遥感]