一种利用参考图象与路面信息的道路监控摄像机标定方法  

A Calibration Method for Road Monitoring Cameras Exploiting Reference Images and Roadway Information

作　　者：张润初 杜倩云 俞祝良[1] 刘浩 张可 

机构地区：[1]华南理工大学自动化科学与工程学院,广东广州510641 [2]北京市交通信息中心,北京100161 [3]北京市交通运行监测调度中心,北京100161

出　　处：《公路交通科技》2014年第11期137-141,共5页Journal of Highway and Transportation Research and Development

基　　金：北京市科技计划项目(Z131106002813012);中国博士后基金项目;北京市博士后工作经费资助项目

摘　　要：鉴于基于视频处理技术的交通流检测系统需要进行摄像机的标定才能得到准确的车流速度信息,而传统的人工标定法无法满足道路监控摄像机发生人为转动时的这一要求,提出了一种基于参考图像与路面信息的的道路监控摄像机标定方法。该方法利用车道标线的宽度和间距以及一条垂直于车道标线的线段,即可估计出摄像机的焦距、俯角、水平偏角和距离地面的高度,同时,为了计算"病态条件"下的摄像机参数,提出了在焦距保持不变的情况下,通过旋转摄像机来获取一组合适的参考图像,再通过参考图像来实现标定。在摄像机发生人为转动的情况下,也能方便地通过参考图像和路面信息来实现重标定。仿真试验表明,该方法具有操作方便、计算简单精确的优点,且无需人工现场操作,大大减少了外业工程量。Aiming at the fact that the video processing based traffic flow detection system needs to calibrate camera to get accurate vehicle speed while the traditional manual calibration methods cannot satisfy this requirement once the camera has been manually rotated, a roadside monitoring camera calibration method is proposed which exploits the reference images and the roadway information. The proposed method only requires two parallel lane markings with known width and a line perpendicular to the lane markings, then the camera parameters such as focal length, tilt angle, pan angle and the camera height can be estimated. Meanwhile, to calculate the ill-condition cameras＇ parameters, a method based on reference images is proposed, the reference images are gained by rotating camera under the condition of keeping the focal length unchanged. It can easily realize cameras re-calibration through reference images and roadway information when the cameras have been rotated manually. The simulation experiment demonstrates that the proposed method has advantages of convenient operation and simple but accurate calculation, in addition, it requires no manual operations and saves much manpower on spot.

关 键 词：交通工程 交通流检测 摄像机标定 车道标线 视频处理 

分 类 号：U491[交通运输工程—交通运输规划与管理]