基于SVM的回归模型路口车辆计数方法  

A Crossroad Vehicle Counting Method Based on SVM-based Regression Model

作　　者：孙耀航 刘萍[1] SUN Yaohang;LIU Ping(Basic Computing Center,Chinese Peoples Liberation Army Academy,Hefei 230031)

机构地区：[1]中国人民解放军陆军军官学院基础部计算中心,合肥230031

出　　处：《计算机与数字工程》2019年第2期446-450,共5页Computer & Digital Engineering

摘　　要：论文提出了一种利用监控视频对路口车辆进行计数的新的方法。针对路口车辆交通拥堵、遮挡程度高、移动速度缓慢、车行有规律等特征,提出基于回归模型的路口监控视频车辆计数算法。首先对监控视频进行处理,从中提取车流特征。之后,利用SVM对实时道路车辆密度进行辨别,并根据辨别结果,对密度高或低的情况分别应用不同的回归模型进行计数。最后,以实地采集的路口交通视频对该方法进行了验证。实验结果表明,该方法可以在实现较高准确率的基础上,占用较少的计算资源,对路口监控视频的车流进行计数。In this article,a novel method of vehicle counting using surveillance video employed on crossing roads is proposed.Since videos on crossing roads tend to present jammed traffic with heavy occlusion,slow movement,and are highly regular,a count-ing algorithm based on regression models is proposed. First the surveillance video to extract features from the video is processed.Then,SVM is used to verify the vehicle density. Finally,the method using real crossing road videos is tested. Experiments showedthat this method is able to count the traffic with higher accuracy and lower computational costs.

关 键 词：SVM 回归模型 车辆计数 

分 类 号：O213[理学—概率论与数理统计]