改进遗传算法优化的温室移动机器人单目相机标定  

Improved genetic algorithm optimization for single camera calibration of greenhouse mobile robots

作　　者：吴镛 陈丰 曹波 苏祥祥 李娜 WU Yong;CHEN Feng;CAO Bo;SU Xiangxiang;LI Na(College of Mechanical Engineering,Anhui Science and Technology University,Fengyang 233100,China)

机构地区：[1]安徽科技学院机械工程学院,安徽凤阳233100

出　　处：《安徽科技学院学报》2024年第3期78-86,共9页Journal of Anhui Science and Technology University

基　　金：安徽省科技特派员农业物质技术装备领域揭榜挂帅项目(2022296906020001);安徽省高校自然科学研究重大项目(KJ2021ZD0110);2022年度合工大智能院“科技成果培育专项”(IMIPY2022019);安徽省自然科学基金面上项目(2308085ME150);安徽省智能农机装备研究院建设协同创新项目(GXXT2023111,GXXT2023110)。

摘　　要：目的:针对温室移动机器人领域传统的张正友相机标定方法存在相机内参数的求解可能陷入局部最优解的问题,使用一种基于改进遗传算法的单目相机内参数优化方案。方法:首先使用张正友标定法得到初始的内参值,然后引入改进的遗传算法对内参初始值进行优化,最后将优化后的内参值代入相机成像模型,得到优化后的重投影误差。结果:将本算法优化后的重投影误差与标准算法和张正友标定法得到的进行比较,证明本研究标定的单目像机内部参数更加准确,标定速度和稳定性更好。结论:改进遗传算法平均重投影误差低于标准算法和张正友标定法,且图像主点坐标更接近参考值,能较好地缓解过早收敛和停滞现象,提高了标定精度,在温室移动机器人视觉应用系统中,本方法有较强的的实用价值。Objective:To solve the problem that the traditional Zhang Zhengyou camera calibration method in the field of greenhouse mobile robots may fall into the local optimal solution,a monocular camera parameter optimization calibration scheme based on improved genetic algorithm was proposed.Methods:Firstly,Zhang Zhengyou's calibration method was ased to obtain the initial internal parameter values,then an improved genetic algorithm was introduced to optimize the initial internal parameter values.Finally,the optimized internal parameter values were substituted into the camera imaging model to obtain the reprojection error.Results:Comparing the reprojection error optimized by the improved algorithm in this article with that obtained by the standard algorithm and Zhang Zhengyou calibration method,this paper demonstrates that the calibrated intrinsic parameters of the monocular camera are more accurate,with improved calibration speed and stability.Conclusion:The average reprojection error of the improved genetic algorithm is lower than that of the standard algorithm and Zhang Zhengyou calibration method,and the main point coordinates of the image are closer to the reference value,which can effectively alleviate premature convergence and stagnation,and improve calibration accuracy.The method proposed has strong practical value in the vision application system of greenhouse mobile robots.

关 键 词：单目相机标定 改进遗传算法 内参优化 重投影误差 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术] TP242.6[自动化与计算机技术—计算机科学与技术]