摘要
针对异质传感器坐标系融合的准确性问题,提出一种利用圆柱面标定物以离线方式计算激光雷达与双目相机的外参数标定方法。通过分析圆柱面点云的三维几何特性,二维化圆柱面点集,并拟合多个平面椭圆实现圆柱面的检测,识别圆柱面图像的棋盘格角点并进行角点三维重建,利用基于邻域法向量的拟合算法拟合圆柱面,并根据最小化投影距离约束条件对激光雷达与双目相机的位姿关系进行计算。仿真与试验结果表明:激光雷达识别圆柱面的半径误差最大为8 mm,双目相机识别圆柱面的半径误差最大为40 mm;且根据外参数标定结果,圆柱面圆度值符合误差特性,这验证了圆柱面标定物应用于激光雷达与双目相机外参数标定的有效性与准确性。
A calibration method of extrinsic parameters of light detection and ranging(LiDAR)and binocular camera is proposed to calculate offline by using a cylindrical calibration surface.With analyzing three-dimensional geometric features of point cloud about cylindrical surface,the dimensionality of the cylindrical point set is firstly reduced,and multiple plane ellipses are fitted to realize the detection of cylindrical surface in point clouds;then the checkerboard corner points of cylindrical image are detected to fit cylindrical surfaces using a fitting algorithm based on neighborhood normal vectors;finally,the attitude relationship(external parameters)between LiDAR and binocular camera is calculated with the constraint of minimizing the projection distance.The simulation and actual experimental results show that the maximum radius error is 8 mm by using LiDAR to detect cylindrical surface,and the maximum radius error is 40 mm by using binocular camera to detect cylindrical surface,and according to the external parameter calibration results,the cylindrical surface roundness value conforms to the error characteristics,which can verify the accuracy about external parameter calibration between LiDAR and binocular camera by using cylindrical surface as calibration object.
作者
张程
刘彬
印曦
戴红良
ZHANG Cheng;LIU Bin;YIN Xi;DAI Hongliang(Shanghai Marine Equipment Research Institute,Shanghai 200031 China;Zhejiang Scientific Research Institute of Transport,Hangzhou 310000,China)
出处
《船舶工程》
CSCD
北大核心
2023年第10期116-125,共10页
Ship Engineering
基金
中国船舶集团自立科技研发专项目(202104Z)
2021年度上海市“科技创新行动计划”社会发展科技攻关项目(21DZ1205800)
浙江省科学技术厅省级重点研发计划(2021C01010)。
关键词
无人船
圆柱面
激光雷达
双目相机
外参数标定
unmanned surface ship
cylindrical surface
light detection and ranging(LiDAR)
binocular camera
external parameter calibration
