一种基于图像处理的超载货车智能监控系统  

An intelligent Surveillance System for Overloaded Truck Based on Image Processing

作　　者：万娇娜 高珊 WAN Jiao-na;GAO Shan(Research Institute of Highway,MOT,Beijing 100088,China;Shandong Traffic Technician College,Shandong 276021,China)

机构地区：[1]交通运输部公路科学研究所,北京100088 [2]山东交通技师学院,山东276021

出　　处：《交通与运输》2018年第A01期193-196,共4页Traffic & Transportation

基　　金：中央级公益性科研所基本科研业务费专项资金项目"面向治超站监管应用的车辆智能视频识别算法研究"(No.2014-9021)

摘　　要：货车超载已经成为公认的安全问题,同时也涉及到运输成本和安全管理人力投入。当前,各国政府都投入了巨资用以治理超载问题。常规方法之一是通过安装视频监控系统来监视道路情况。然而,对于多个卸载场,同时有几十甚至上百监控画面的情况,采用人工观看这种传统的监控模式,效率相对偏低。因此,开发了一种基于图像处理算法的智能超载卡车视频监控系统。该系统采用图像处理算法来自动检测超载管理站的卡车,并定时比对同一位置货车影像以确定是否为同一车辆长时期停在卸载场,未进行卸载处理。通过处理2万多张图片,建立了正负样本数据库。利用Haar-like特征和Adaboost训练算法,从实时监控视频数据流中检测卡车。此外,利用样本图像灰度值分布计算的统计规律,对货车检测结果进行了改进。实际测试验证了系统的有效性。Overloading has been recognized to be both a safety concern as well as a cost concern.The governments of all over the world all make huge investment to manage the overloading phenomena.One of the normal ways is to have an eye on the road circumstances by installing video surveillance system.However,the traditional way,which has some officers watching the screen all day long,is not efficient at all.Therefore,we have developed an intelligent Overload Truck Surveillance system.This system employs image processing algorithms to detect the truck in the overloading managing stations,and compare the van images in the same location to make sure whether it parks in uninstall field for a long time,never unloading.We have built positive and negative samples database by handling more than 20,000 images.Haar-like features and Adaboost training algorithm are used to detect truck from the real-time surveillance video data stream.Moreover,the statistic rules calculated from the grey value distribution of the sample images are employed to refine the truck detection results.This paper shows some interfaces when the system is running,and the efficiency of the system is demonstrated.

关 键 词：道路安全 智能视频监控 图像处理 超载货车 

分 类 号：U491[交通运输工程—交通运输规划与管理]