交通监控场景下的相机标定与车辆速度测量.PDF

37 8 Vol. 37 No. 8 第 卷第 期 计算机应用研究 录用定稿 Application Research of Computers Accepted Paper 交通监控场景下的相机标定与车辆速度测量* 武非凡，宋焕生，戴 喆，王 伟 (长安大学 信息工程学院, 西安 710064) 摘 要：针对交通监控场景中对车辆速度测量的需求，提出了一种相机标定方法和车辆速度测量方案。首先，通过 深度学习YOLO 检测算法和光流跟踪算法对图像中的车辆目标进行检测和跟踪，根据获得的轨迹集合使用级联霍夫 变换计算出道路方向上的消失点，从而检测出道路上的标志线。之后根据消失点和标志线，使用试探焦距思想完成 相机标定任务。最后通过计算多帧之间瞬时速度的平均值来实现车辆速度的测量。通过真实交通监控场景的实验结 果表明，这种基于消失点的自动相机标定方法具有较好的稳定性和较高的标定精度，能够满足车辆速度测量和实际 工程应用的需求。 关键词：相机标定；车辆速度；消失点；标志线；交通监控 中图分类号：TP391.4 doi: 10.19734/j.issn.1001-3695.2019.03.0089 Camera calibration and vehicle speed measurement in traffic monitoring scene Wu Feifan, Song Huansheng, Dai Zhe, Wang Wei (School of Information Engineering , Chang’an University, Xi ’an 710064, China) Abstract: Aiming at the demand for vehicle speed measurement in traffic monitoring scenarios, this paper proposes a camera calibration method and vehicle speed measurement scheme. Firstly, this method detected and tracked the vehicle target in the video based on deep learning YOLO detection algorithm and optical flow tracking algorithm. According to the obtained trajectory set, this paper calculated the vanishing points in the road direction by Cascade Hough Transform, and detected marking lines. Then, according to the vanishing point and marking line, the camera calibration task is completed with the idea of exploratory focal length . Finally, this paper achieved vehicle speed by calculating the average of the instantaneous speeds between the multiple