基于图像的铁路货车车号定位与识别*

摘要

针对铁路货车车号特征，提出一种基于图像处理的车号定位与识别方法。该方法首先对车号区域图像进行预处理并粗定位出车号区域，然后采用改进的 Soble算子边缘检测技术与投影法和铁路货车车号先验知识相结合对车号区域进行精确定位，最后通过对字符进行分割和归一化处理，在粗网格特征提取字符特征的基础上采用模板匹配法快速实现了车号识别。实验结果表明，该方法对铁路货车各种车型车号区域定位和识别具有较高的准确率。%Aiming at the features of freight wagon,a new method,which is based on image process-ing,is proposed to locate and identify train number.Firstly,the number region is pretreated in or-der to roughly locate the region.Then three are combined to locate the region accurately which are the edge detection technique with modified Soble operator,proj ection method and prior knowledge of railway train number.Finally,through character division and normalization procession,tem-plate matching method is used to recognize the number and type of the freight wagon as fast as possible,which is based on character feature extraction on coarse mesh.The experimental results show that the new method performs well in accurately locating and identifying the train number.