基于视频技术的交通流量预测

Traffic flow prediction based on video technology

城市交通拥堵是一个世界性的难题,至今还没有特别有效的解决方法。目前正在研究、实施的智能运输交通系统(ITS),主要是通过对交通流量的合理调控,来优化路网的利用率。基于各路口的视频摄像系统,通过采集卡将获取的图像转化为实际的车流量信息。根据人工神经网络的理论建立BP算法的流量预测模型,用实际测得的流量信息,在MATLAB环境下对模型进行训练。然后再利用训练好的模型来预测路口将来几个时段的交通流量。如果预测到某方向的流量将很大,可以通过实时修改信号灯的软件参数,延长该方向的绿灯时间,化解可能出现的交通拥堵。

Urban traffic jam is a worldwide problem, which has not yet been solved effectively. The world＇s major developed countries are studying the implementation of Intelligent Transport Systems （ITS）, which mainly optimizes network utilization rate through the rational control of traffic flow. the images will trans- fered into the actual traffic flow information based on the intersection of video camera system through the acquisition card. Traffic prediction model of BP algorithm is established according to the theory of artificial neural network. The model is trained through using the actual measured flow information in the MATLAB environment. Ttraffic flow of several periods is predicted with the trained model. If the predicted dates are larger, the the software parameters of signal lamp in real time will be modified to extend the time of green light and resolve the traffic problems.