一种实时路况下的车牌照识别系统

A Real-time Vehicle License Plate Recognition System

该系统采用高清工业相机，以千兆以太网作为数据传输通道，克服了传统传输方式速度较慢，抗干扰能力差的缺点。在识别方面，提出了自适应模板宽度算法，有效克服了光照对车牌定位的影响。该算法还采用投影分割法精确分割车牌字符，对汉字采用Gabor算子提取特征，然后利用SVM进行识别，有效解决了汉字识别问题。

In this paper,the authors use a high-definition industrial camera for image acquisition and Gigabit Ethernet for data transmission to overcome the shortcomings of traditional slow transmission rate and the poor anti-interference ability. A novel algorithm of self-adaptive adjustment of template width is proposed, which effectively decreases the influence of light in locating the license plate in image. By way of projection,the system can partition Chinese characters in license plate accurately. After ex- traction of their features by Gabor operator, the authors use SVM to identify them. The experiments show that it is an effective method to solve the problem of Chinese character recognition.