DOI: 10.11908/j.issn.0253-374x.19543

作者:

作者单位: 1.同济大学 软件学院，上海 201804;2.印第安纳大学与普渡大学印第安纳波利斯联合分校，印第安纳波利斯 46202;3.同济大学 道路与交通工程教育部重点实验室，上海 201804


作者简介: 高 珍（1979—），女，副教授，工学博士，主要研究方向为数据分析与数据挖掘。E-mail:gaozhen@tongji.edu.cn


通讯作者:

中图分类号: TP311;TP391.


基金项目: 上海市科学技术委员会(18DZ1200200)；国家自然科学基金(51878498)；上海一流研究生教育建设项目（ZD19040605）




Real-Time Detection Algorithmof Traffic FlowParameters Based on Sequence Diagram of Vehicle Square Wave Pulse
Author:

Affiliation: 1.School of Software Engineering, Tongji University, Shanghai 201804, China;2.Indiana University Purdue University Indianapolis, Indianapolis 46202, USA;3.Key Laboratory of Road and Traffic Engineering of the Ministry of Education, Tongji University,Shanghai 201804, China


Fund Project:




摘要
| 图/表
| 访问统计
| 参考文献
|相似文献
| 引证文献
| 资源附件

摘要:提出一种基于车辆方波脉冲时序图的交通流参数实时检测算法，克服了现有方法易受光线变化及天气影响、运算量大等缺陷，提高了实时交通流参数检测的准确率，为智能交通系统提供有力支撑。研究基于虚拟检测线，将交通监控视频降维处理为包含时间和空间信息的时空图，而后对时空图进行前景提取，生成二值化时空图的垂直投影，针对像素累积图设计了系统性去噪及车辆对象识别方法，进而生成车辆方波脉冲时序图，实时检测出车流量、车头时距、时间占有率、车辆速度并进行车辆分类。分析结果表明，所提出的方法能克服雨雪天气、夜晚光线等干扰，快速而准确地进行多车道交通流参数获取，计算负荷小，方法准确率高达97.32%，可满足智能交通系统对交通流参数检测实时性和精度的要求。



Abstract:The video-based traffic surveillance is widely studied nowadays. But the existing methods are always challenged by the influence of light changing, weather effects, and a large amount of computation. This paper proposed a novel method of using sequence diagram of vehicle square wave pulse to process and analyze road monitoring videos based on spatial-temporal profileaiming at providing real-time detection of traffic flow parameters and vehicle classification for intelligent transportation system (ITS). First, based on the setting of virtual detection line, this method reduces a large number of traffic monitoring videos into spatial-temporal profiles, which contain time and space information. Next, the foreground of the spatial-temporal profile is extracted to generate a vertical-projected pixel histogram. Finally, vehicle objects are detected and the traffic state parameters are calculated, including traffic flow, time headway, occupancy, vehicle speed, and vehicle classification. The analysis result shows that the proposed method can obtain traffic flow parameters quickly and accurately even with the interference of weather and light. The accuracy rate of the method is as high as 97.32%, which is efficient and practicable to satisfy the real-timeand accuracy requirements of detection of traffic flow parametersin intelligent transportation systems.





PDF全文下载地址:

点我下载PDF