基于视频检测的高速公路交通事件影响预测研究

发布时间：2018-11-23 11:20

【摘要】：随着城市的发展及城市间交通需求的急剧增长,城间高速公路在高峰期拥挤、阻塞日益严重,交通事件频繁发生。针对高速公路交通事件容易引发大规模拥堵等问题,通过有效的事件检测技术可及时针对事件进行处理,减少二次事件等不良反应的发生；而针对事件影响范围进行实时预测分析,可以作为制定事件解决方案的依据。本文主要针对高速路事件检测与影响预测方法进行设计研究。本文在广泛调研国内外高速路事件检测和影响预测算法研究现状的基础上,提出一种基于视频的快速事件检测算法,并通过分布式视频检测信息对事件影响范围进行动态预估,实现事件信息的快速反馈。首先,利用平稳序列法获取背景,由背景差法获取前景,并在使用凸包占有率对遮挡进行检测的基础上,采用改进的卡尔曼滤波特征匹配跟踪算法对车辆进行跟踪；然后通过对交通流的速度和流量进行检测,建立速度、流量与交通流状态之间的映射关系,运用模糊推理方法判别交通事件的发生；最后,以交通事件车流波模型作为理论依据,通过分布式视频信息获取实时排队长度与车流波速度,建立未来多个时刻的车辆排队动态估计模型,针对交通事件的影响范围进行预测分析。综上所述,本文先以事件检测算法、排队检测算法和车流波理论为基础,然后结合计算机视觉和图像处理的相关知识,通过摄像机拍摄到的道路交通流信息,检测交通事件的发生,并基于分布视频信息计算车流波速度,对事件影响范围进行预测,最后完成事件检测与影响预测系统的设计与实现。实验表明,本文提出的事件检测方法能准确地获取前景信息,快速有效地对高速公路上的交通事件进行检测。而影响预测方法较其他方法更加准确有效,能够作为高速公路交通管理人员调度和决策的依据。
[Abstract]:With the development of cities and the rapid growth of intercity traffic demand, intercity highways are congested and congested in the rush hour, and traffic events occur frequently. In view of the problem that expressway traffic events are easy to cause large-scale congestion, the effective event detection technology can deal with the events in time and reduce the occurrence of adverse reactions such as secondary events. The real-time prediction and analysis of the impact range of events can be used as the basis for the formulation of event solutions. This paper focuses on the design and research of highway event detection and impact prediction methods. On the basis of extensive investigation of the current situation of highway event detection and impact prediction algorithms at home and abroad, this paper proposes a fast event detection algorithm based on video, and dynamically estimates the impact range of events through distributed video detection information. The fast feedback of event information is realized. Firstly, the background is obtained by stationary sequence method, the foreground is obtained by background difference method, and the improved Kalman filter feature matching tracking algorithm is used to track the vehicle on the basis of using convex hull occupation rate to detect occlusion. Then, by detecting the speed and flow of traffic flow, the mapping relationship between speed, flow and traffic flow state is established, and the fuzzy inference method is used to distinguish the occurrence of traffic events. Finally, based on the traffic event vehicle flow wave model, the real-time queue length and the vehicle flow wave velocity are obtained by distributed video information, and the vehicle queue dynamic estimation model at many times in the future is established. Forecast and analyze the influence range of traffic events. To sum up, this paper first based on event detection algorithm, queue detection algorithm and vehicle flow wave theory, and then combined with computer vision and image processing related knowledge, through the camera captured by the road traffic flow information, The traffic events are detected, and the vehicle flow velocity is calculated based on the distributed video information, and the influence range of the events is predicted. Finally, the design and implementation of the event detection and impact prediction system are completed. The experimental results show that the proposed event detection method can obtain the foreground information accurately and detect the traffic events on the freeway quickly and effectively. The influence prediction method is more accurate and effective than other methods, and can be used as the basis for highway traffic management personnel scheduling and decision-making.
【学位授予单位】：山东大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U495;TP391.41

【相似文献】