基于多源信息融合的交通要素识别方法研究

车型识别是智能交通系统的重要组成部分。针对特定类车辆脸部特征相近,提取车头特征易导致模型辨别力差、识别精度低等问题。提出了一种基于车辆侧面特征的车型识别方法。采用卷积神经网络实现对不同类型车辆的检测,使用统计模型计算目标车辆的横向位置。建立双相机(触发相机和抓拍相机)协同跟踪模型,利用感知哈希算法,对目标车辆实现判别式跟踪。最后抓拍相机完成车辆正面抓拍,完成目标车辆的车牌识别。最终的检测实验结果取得了81.94%的平均正确均值(Mean Average Precision,mAP)。

Research on Traffic Factor Recognition Based on Multi-source Information Fusion

Vehicle type recognition is an important part of ITS.Due to the similar facial features of a certain type of vehicles,extracting the front feature of the vehicle can easily lead to the problem of insufficient expression capability and low recognition accuracy.A vehicle type recognition method based on vehicle side features is proposed.The convolutional neural network is used to detect different types of vehicles,and the statistical model is used to predict the driving lanes.The dual camera(trigger camera and snap camera)cooperative tracking model is established,and the Perceptual Hash Algorithm is used to realize the discriminant tracking of the target vehicle.Finally snap camera captures the front of the vehicle to complete the target vehicle license plate recognition.The final experimental result achieved an mean average precision(mAP)of 81.94%.