基于尾灯跟踪的夜间车辆检测.pdfVIP

2012年第10期，第 45卷 通 信 技 术 Vol.45，No.10,2012 总第250期 Communications Technology No.250,Totally 基于尾灯跟踪的夜间车辆检测 祁秋红， 陈启兴 （成都信息工程学院 通信工程学院，四川 成都 610225） 【摘 要】夜间车辆检测在城市交通检测技术中有着重要的研究意义。夜晚尾灯的彩色信息作为车辆的 一个显著标志，能帮助我们比较准确地检测到车辆。因此，提出了一种基于尾灯跟踪的夜间车辆检测方法。 先通过 HSV 颜色模型对尾灯的颜色信息进行分割，辨认出车辆位置，再通过区域边界锁定车灯边缘信息对车 辆进行跟踪。实验证明，该方法提高了检测效率，减少了白色路灯干扰，比较有效地检测到行驶车辆。 【关键词】车辆检测；HSV 颜色空间；尾灯检测；边界区域 【中图分类号】TP391.41 【文献标识码】A 【文章编号】1002-0802(2012)10-0058-03 Vehicle Detection based on Two-way Multilane at Night QI Qiu-hong， CHEN Qi-xing (College of Communication Engineering, Chengdu University of Information and Technology, Chengdu Sichuan 610225, China) 【Abstract】Vehicle detection at night plays an important role in urban traffic detection technology. The color information of tail lights, as a significant sign of vehicle at night is helpful to the exact detection of the vehicles. This paper presents a method to detect vehicles by tracking the tail light. The tail-light color information is first segmented according to the HSV color space thus to locate the vehicle, then a bounding box is set to detect the vehicle by area boundary of the tail light. Practice indecates that this method could improve the detection efficiency, reduce the interference of white streetlights, and thus effectively detect the running vehicle. 【Key words】vehicle detection; HSV color space; tail-light detection; bounding box 0 引言 邻帧差和隔帧帧差想结合的累积帧差法来检测夜间 日益增加的机动车辆，给城市交通带来了负担，