一种多摄像头车辆实时跟踪系统  

A real time vehicle tracking system with multiple cameras

作　　者：崔瑞 贾子彦[1] Cui Rui;Jia Ziyan(Jiangsu University of Technology,Changzhou 213000,China)

机构地区：[1]江苏理工学院,江苏常州213000

出　　处：《无线互联科技》2023年第14期39-42,共4页Wireless Internet Technology

摘　　要：随着城市人口的增加,越来越多的车辆使得城市的交通状况越来越复杂。针对传统的车辆检测中出现的跟踪车辆易丢失、跟踪精度低等问题,文章提出一种基于多摄像头的车辆实时跟踪检测方法,从多角度对运动车辆进行跟踪。在分析YOLOv5算法的基础结构后,文章针对车辆尺度变化大的特点,充分利用YOLOv5算法检测轻量化、速度快、实时性强的性质,并在此基础上利用多个摄像头之间的单应性矩阵来确定车辆位置的变化。结合颜色特征识别和车辆特征识别对车辆进行重识别,不仅提高了运行速度,而且满足了实时性和准确性的要求,有效解决跟踪车辆易丢失的问题,获得了较为成功的车辆实时跟踪效果。With the increase of urban population year by year,more and more vehicles make urban traffic conditions more and more complicated.Aiming at the problems of tracking vehicles easy to lose,low tracking rate and poor real-time performance in traditional vehicle detection,a multi-camera based vehicle real-time tracking detection method is proposed to track moving vehicles.After analyzing the basic structure of the YOLOv5 algorithm,and considering the large scale variation of vehicles,the YOLOv5 algorithm is fully utilized to detect the properties of lightweight,fast speed and strong real-time.On this basis,the homography matrix between multiple cameras is used to determine the changes of vehicle position.Combining color feature recognition and vehicle feature recognition,vehicle rerecognition not only improves the running speed,but also meets the requirements of real-time and accuracy,effectively solves the problem that tracking vehicles are easy to lose,and finally achieves a more successful real-time vehicle tracking effect.

关 键 词：多摄像头 YOLOv5 单应性矩阵 特征识别 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]