Abstract: A real-time trucks recognition method is put forward for heavy traffic flow multi-lane roads which are no entry for trucks.Firstly，big vehicles are detected by the color 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 license plates in the close shot and the panorama is relative fixed.Secondly，based on mean shift segmentation and edge detection，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 efficiency，which has high practical application value.

Key words: intelligent transportation system, truck recognition, no entry, mean shift, fuzzy rules

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mean shift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。

关键词: 智能交通系统, 大货车识别, 禁行, mean shift算法, 模糊规则