基于视频的车辆检测在交通噪声预测中的应用

Application of Video-Based Vehicle Detection in Traffic Noise Prediction

目前交通噪声预测使用的FHWA模型,在实际应用中以人工方式来统计通过的车辆数和车型,此法成本高,易产生偏差。通过引入检测线上发生改变像素的连接程度,提出基于中介度量理论和车辆长度信息的车辆检测算法,自动获得道路通过的车辆数和车型,车辆检测结果可与FHWA预测模型有效结合。实验结果表明,算法达到了很好的效果,为FHWA模型提供了准确的车流量数据。

At present,the number and type of vehicles are counted by hand in traffic noise prediction by means of the FHWA model.The cost of labor is high and the statistical value often has low accuracy.By introducing the connection level of changed pixels on the virtual detection line,a new vehicle detection algorithm is proposed based on the theory of medium measurement logic and the information of the vehicle length.The algorithm can count the number and type of vehicles automatically.The result of detection can combine with the FHWA model effectively.The experiment results show that the algorithm has a good performance,and can provide accurate traffic volume data for the FHWA model.