用于工业级的远距离相机手眼两步标定方法  

A Two-step Hand-eye Calibration Method for Industrial-grade Long-Distance Cameras

作　　者：潘章斌 潘海鸿[2] 莫玉良 陈家春 陈琳[2] PAN Zhangbin;PAN Haihong;MO Yuliang;CHEN Jiachun;CHEN Lin(Guangxi Natural Resources Vocational And Technical College,Nanning,Guangxi 532199,China;College of Mechanical Engineering,Guangxi University,Nanning,Guangxi 530004,China)

机构地区：[1]广西自然资源职业技术学院,南宁532199 [2]广西大学机械工程学院,南宁530004

出　　处：《自动化与仪器仪表》2025年第1期156-159,共4页Automation & Instrumentation

基　　金：广西创新驱动发展专项(桂科AA18118002)。

摘　　要：为解决工业应用中远距离的相机手眼标定中所存在的标定误差大、标定步骤繁琐等问题,提出一种远距离相机手眼两步标定方法。该方法将标定板固定于机器人六轴,避免标定板成像距离过远而导致的标定误差增大的问题;基于机器人世界坐标系规划机器人平移运动,仅需三次机器人平移操作即可快速标定机器人手眼矩阵;以角点坐标系构建超定方程,并基于高斯-牛顿方法对手眼矩阵进行非线性优化,有效降低手眼矩阵的非线性误差。实验结果表明,该方法能有效完成远距离相机手眼矩阵标定,其视觉定位的平均欧式误差和最大欧式误差相较于九点标定法分别下降48.099%和49.688%,能有效保证手眼矩阵标定精度。A long-distance camera hand-eye two-step calibration method was proposed to solve the accuracy problems in industrial applications.This method fixes the calibration plate on the six axes of the robot to avoid the problem of increasing the calibration error caused by the imaging distance of the calibration plate being too far.The robot's translational motion is planned based on the robot's world coordinate system,and the robot's hand-eye matrix can be quickly calibrated with only three robot translational operations.Finally,an overdetermined equation is constructed in the corner coordinate system,and the hand-eye matrix is nonlinearly optimized based on the Gauss-Newton method,which effectively reduces the nonlinear error of the hand-eye matrix.The experimental results show that this method can effectively complete the long-distance camera hand-eye matrix calibration.Compared with the nine-point calibration method,the average Euclidean error and the maximum Euclidean error of visual positioning are reduced by 48.099%and 49.688%respectively,which can effectively ensure the accuracy of the hand-eye matrix calibration.

关 键 词：远距离成像 快速标定 手眼标定 非线性优化 

分 类 号：TP394.1[自动化与计算机技术—计算机应用技术] TH691.9[自动化与计算机技术—计算机科学与技术]