大货车闯禁行的实时识别方法  

Real-time trucks recognition method for truck-no-entry roads

作　　者：磨少清[1,2] 刘正光[1] 张军[1] 

机构地区：[1]天津大学电气与自动化工程学院,天津300072 [2]广西民族大学物理与电子工程学院,南宁530006

出　　处：《计算机工程与应用》2010年第22期225-228,共4页Computer Engineering and Applications

基　　金：天津市公安交通局科研基金(公科[2005]16号)

摘　　要：针对车流量大的多车道路面,提出了一种大货车闯禁行实时智能识别方法。该方法首先根据车牌颜色检测大车,进而以车牌在协同拍摄的远近景中存在相对固定的位置关系通过拟合函数及局部搜索技术获取大车,然后提出了基于meanshift分割与边缘检测相结合的方法以获取3个能有效区分大客车和大货车的结构特征,最后使用模糊规则实现大货车的识别。实测结果表明提出的方法能有效克服光照和模糊等干扰,准确率高,识别速度快,能满足实时性要求,具有较高的应用价值。A real-time trucks recognition method is put for trucks.Firstly, big vehicles are detected by the color forward for heavy traffic flow multi-lane roads which are no entry of license plate, and then is segmented from the panorama by the combination of a mapping function and local searching which is stimulated by the fact that the position relationship of li- cense plates in the close shot and the panorama is relative fixed.Secondly,based on mean shift segmentation and edge detec- tion, a method for extracting three structural features which can distinguish trucks from buses efficiently is proposed.At last, trucks are recognized by the fuzzy rule.Field experimental results indicate the proposed method is robustness,accuracy and ef- ficiency,which has high practical application value.

关 键 词：智能交通系统 大货车识别 禁行 mean SHIFT算法 模糊规则 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]