一种新的基于形态学和模板匹配的车牌识别方法

New license plate recognition based on morphology and template matching

信息技术的发展极大地推动了智能交通的进程,研究实现实时、准确、高效的车牌识别系统,是当前智能交通领域的重要挑战之一。针对传统车牌定位和识别方法存在的不足,研究提出一种车牌定位识别的新方法：首先通过新的形态处理和窗口搜索,解决定位错误或出现多个候选区域和在复杂背景下可能出现漏检车牌的问题,找出类似的车牌区,并根据车牌的长宽比去除假车牌区,来进行车牌准确定位;然后结合模板匹配新的匹配算法,来进行车牌字符识别。实验结果表明：相比其他算法该算法实现简单;不需要很多样本进行训练;定位精度较高;反应时间较快,小于4s;能够较好的解决定位失败和车牌漏检的问题,基本满足实时性的要求。

The development of information technology has greatly promoted the process of intelligent transportation.It is one of the important challenges in the field of intelligent transportation system to study the real-time,accurate and efficient license plate recognition system.A new method of license plate location recognition is proposed to solve problems of traditional methods：Firstly,a new approach of window search and morphology process is presented to overcome possible problems error positioning at multiple candidate area,and identify similar license plate area on the aspect ratio of the number plate;In the end,a license plate recognition is completed by a new matching algorithm proposed here.Experimental results show that compared with other algorithms,the algorithm is simple,high positioning accuracy,and less than 4 s.Especially,the algorithm don＇t need a lot of training samples,may better solve the problem of locating failure and license plate missing,and basically meet the requirement of real-time.