使用安置参数优化的无人机点云与影像精配准  

Precise registration of UAV point clouds and images using placement parameter optimization

作　　者：刘波[1] 邹鹏辉 刘华 LIU Bo;ZOU Penghui;LIU Hua(Faculty of Geomatics,East China University of Technology,Nanchang 330013,China)

机构地区：[1]东华理工大学测绘工程学院,南昌330013

出　　处：《测绘科学》2021年第9期94-101,108,共9页Science of Surveying and Mapping

基　　金：国家自然科学基金项目(42001411);江西省自然科学基金项目(20202BABL212014);东华理工大学江西省数字国土重点实验室开放研究基金资助项目(DLLJ202004)。

摘　　要：针对现有的无人机激光点云与影像配准方法主要基于仿射变换建立点云与影像之间的配准模型,无法描述激光点云与影像之间的复杂变换关系问题,该文提出一种基于安置参数优化的无人机激光点云与光学影像精确配准方法。以相机安置参数为优化参数,以归一化互信息作为点云特征影像与光学影像之间的相似性测度,采用改进Powell算法作为优化策略,获得最优配准参数,实现无人机激光点云与无人机影像之间的精确配准。基于实验数据与现有两种配准方法进行精度对比。实验表明,基于相机安置参数优化的配准方法优于两种常规配准方法,可以实现无人机激光点云与光学影像之间的自动化配准,并达到像素级的配准精度。To solve the problem that the existing registration methods for UAV(Unmanned Aerial Vehicle)laser point clouds and images were mainly based on affine transformation to establish the registration relationship between point clouds and images and could not fully describe the complex transformation between laser point clouds and images,this paper proposed an accurate registration method of UAV laser point clouds and optical images based on the optimization of placement parameters.In order to achieve accurate registration result,camera placement parameters were taken as the optimization parameters,normalized mutual information was taken as the similarity measure between point cloud feature images and optical images,and improved Powell algorithm was used as the optimization strategy to obtain the optimal registration parameters.Based on the experimental data,the experiment was compared with the existing two registration methods.The experimental results showed that the registration method based on the optimization of camera placement parameters was better than the two conventional registration methods,which could realize the automatic registration between UAV laser point cloud and optical image,and achieved the pixel level registration accuracy.

关 键 词：无人机激光点云 点云影像配准 安置参数 归一化互信息 改进Powell算法 

分 类 号：P237[天文地球—摄影测量与遥感]