当前位置:主页 > 科技论文 > 软件论文 >

基于单目视觉的GPS辅助相机外参数标定

发布时间:2018-10-11 08:58
【摘要】:针对机器人视觉系统外参数标定的问题,提出了基于单目视觉ORB-SLAM的差分GPS辅助相机外参数标定方法。分析了单目视觉ORB-SLAM和GPS(Global Position System)定位数据之间的相似关系,建立了相机外参数标定的非线性最小二乘模型。基于随机采样一致性(RANSAC),通过三点法求得模型的初始解。设计了Levenberg-Marquardt(LM)迭代算法求解出最优解,从而得到了最优的相机相对位置和姿态参数。最后,对提出的方法进行仿真和跑车试验验证。结果表明:在试验半径为50m时,所设计标定方法的姿态标定精度可达0.1°,位置标定精度可达0.2%。该方法标定过程简单实用,不需要外界环境的先验信息和人工干预,具有很高的精度和显著的应用价值。
[Abstract]:Aiming at the problem of external parameter calibration of robot vision system, a method of external parameter calibration for differential GPS camera based on monocular vision ORB-SLAM is proposed. The similarity relationship between Monocular Vision ORB-SLAM and GPS (Global Position System) positioning data is analyzed, and a nonlinear least square model for camera external parameter calibration is established. The initial solution of the model is obtained by three point method based on random sampling consistent (RANSAC),. The Levenberg-Marquardt (LM) iterative algorithm is designed to solve the optimal solution, and the optimal camera relative position and attitude parameters are obtained. Finally, the proposed method is verified by simulation and sports car test. The results show that when the experimental radius is 50 m, the precision of attitude calibration can reach 0.1 掳, and the precision of position calibration can reach 0.2 掳. The calibration process of this method is simple and practical. It does not require prior information and manual intervention from outside environment. It has high accuracy and significant application value.
【作者单位】: 海军航空工程学院控制工程系;海军航空兵学院空中领航系;
【基金】:国家自然科学基金资助项目(No.61004002)
【分类号】:TP242;TP391.9

【相似文献】

相关期刊论文 前10条

1 张俊;闫镔;李磊;陆利忠;张峰;;锥束CT系统旋转平移轨迹几何参数标定方法[J];强激光与粒子束;2013年10期

2 张俊;闫镔;陆利忠;李磊;张峰;;一种锥束CT系统几何参数标定方法[J];核电子学与探测技术;2014年02期

3 黄立娜;余仲秋;;电子白板系统中参数标定方法的研究[J];应用光学;2012年02期

4 李凌e,

本文编号:2263604


资料下载
论文发表

本文链接:https://www.wllwen.com/kejilunwen/ruanjiangongchenglunwen/2263604.html


Copyright(c)文论论文网All Rights Reserved | 网站地图 |

版权申明:资料由用户74e49***提供,本站仅收录摘要或目录,作者需要删除请E-mail邮箱bigeng88@qq.com