基于视频与检测器数据组成的语义层次交通事件检测

Incident Detection Based on Video and Detector Data Composed of Semantic Hierarchy

车辆跟踪中普遍存在车辆遮挡将直接影响事件检测精度的情况。为了解决这一问题,在介绍车辆跟踪算法基本原理的基础上,提出了一种基于视频与检测器数据组成的语义层次交通事件检测算法。该算法先应用时空马尔可夫随机场模型进行车辆跟踪,得到交通流基本参数,然后结合安装在道路下游的检测器获得的交通流数据一起,采用语义层次算法对交通事件进行检测。为了验证算法的准确性,最后对该算法与不使用检测器数据算法进行比较,发现使用检测器数据算法的检测率要高。通过研究得出:基于视频与检测器数据组成的语义层次算法在交通量比较拥挤且车辆出现相互遮挡的情况下,能准确判断交通事件发生,对缓解交通拥挤和减少交通事故有重要的意义。

As the common problem in vehicle tracking, vehicle occlusion directly impacts on the incident detection precision. In order to solve the problem, on the basis of the basic principle of vehicle tracking algorithm, a traffic incident detection algorithm based on video and detector data composed of semantic hierarchy is proposed. The algorithm applies spatio-temporal Markov random field model to track vehicle and obtain traffic flow~ basic parameters, then combines the data of traffic flow from downstream road detectors, at last uses the semantic hierarchy algorithm to detect traffic incident. In order to verify the accuracy of the algorithm, the proposed algorithm is compared with the algorithm only using video data, which shows that the detection rate of the algorithm using detector data is higher. The result shows that the proposed algorithm can accurately identify traffic event, when the traffic volume is large and vehicle appear mutual occlusion, it is meaningful to ease traffic congestion and reduce number of traffic accidents.