摘要
文中设计了一种基于深度相机的船体曲板测量系统.该系统将深度相机与伺服驱动结合,使用计算机控制采集过程,获取指定位置的多组小范围点云.采用一种超体素聚类改进的区域生长分割方法从原始点云数据中提取曲板表面点云,通过相机安装的相对位置和伺服系统的返回参数值拼接完整的船体曲板三维点云.使用该系统在3 000 mm×3 000 mm范围测量船体曲板,所得非边缘点云平均误差为0.61 mm,符合预期精度要求.
A measurement system of hull curved plate based on depth camera was designed.The system combined depth camera with servo drive,and used computer to control the acquisition process to obtain multiple groups of small-range point clouds at designated positions.An improved region growing segmentation method based on hyper-voxel clustering was used to extract the surface point cloud of curved plate from the original point cloud data.Through the relative position of camera installation and the return parameter value of servo system,the complete three-dimensional point cloud of hull curved plate was spliced.The system was used to measure the curved hull plate in the range of 3000 mm×3000 mm.The average error of non-edge point cloud is 0.61mm,which meets the expected accuracy requirements.
作者
张方哲
于洋
俞峰
胡勇
ZHANG Fangzhe;YU Yang;YU Feng;HU Yong(Wuhan University of Technology,Wuhan 430063,China;Jiangnan Shipyard(Group)Co.Ltd.,Shanghai 201913,China)
出处
《武汉理工大学学报（交通科学与工程版）》
2024年第2期298-303,308,共7页
Journal of Wuhan University of Technology（Transportation Science & Engineering）
基金
国防基础科研项目(JCKY2019206B001)。
关键词
船体曲板
深度相机
三维测量
计算机视觉
点云处理
curved hull plate
depth camera
3d measurement
computer vision
point cloud processing
