计量与测试技术2024,Vol.51Issue(5) :54-58.DOI:10.15988/j.cnki.1004-6941.2024.5.015

一种面向船体扫描的多站点云鲁棒配准技术

A Multi-site Cloud Robust Registration Technique for Hull Scanning

胡敏捷 杨荣淇 谢洪 马琦
计量与测试技术2024,Vol.51Issue(5) :54-58.DOI:10.15988/j.cnki.1004-6941.2024.5.015

一种面向船体扫描的多站点云鲁棒配准技术

A Multi-site Cloud Robust Registration Technique for Hull Scanning

胡敏捷 1杨荣淇 1谢洪 2马琦1
扫码查看

作者信息

  • 1. 上海船舶研究设计院
  • 2. 武汉大学
  • 折叠

摘要

为实现船舱外壳的点云配准,本文提出了一种基于回环约束的多站点云配准方法.首先,使用ISS从场景中提取关键点,将颜色信息与形状信息的直方图相融合,得到船舱点云外壳的CSHOT特征描述符;然后,基于回环的粗配准,构建全连接图,并利用GROR,结合最大一致性集约束和闭环约束移除误匹配;最后,采用全局的精配准LUM算法完成位姿优化.实验表明:该方法的平均RMSE为0.0077m,对船舱外壳的点云配准取结果较好.

Abstract

In order to realize point cloud registration of cabin shell,a multi-site cloud registration method based on loop constraint is proposed in this paper.Firstly,ISS is used to extract key points from the scene,and the histo-gram of color information and shape information is fused to obtain the CSHOT feature descriptor of the cabin point cloud shell.Then,based on the rough registration of the loop,the fully connected graph is constructed,and GROR is used to remove the mismatching by combining the maximum consistent set constraint and the closed-loop con-straint.Finally,the global LUM algorithm is used to complete the pose optimization.The experiments show that the average RMSE of this method is 0.0077 m,and the result is good for the point cloud registration of the cabin shell.

关键词

船舱外壳/CSHOT/最大一致性集约束/闭环约束/多站点云配准

Key words

cabin shell/CSHOT/maximum consistency set constraint/closed loop constraints/multi site cloud regis-tration

引用本文复制引用

出版年

2024
计量与测试技术
成都市计量监督检定测试所

计量与测试技术

影响因子:0.175
ISSN:1004-6941
段落导航相关论文