车牌自动识别系统的研究与实现

为了缓解日益严重的交通压力、提高城市交通管理的工作效率和增强人们的安全防范意识,采用基于纹理和颜色信息的综合车牌定位方法对车牌区域精确定位。通过分析水平投影的统计特征和竖直投影的特征进行车牌字符分割,利用自适应性和学习能力强的BP神经网路进行字符识别,研究和实现了车牌自动识别系统。测试结果显示,车牌识别时间小于200 ms,识别率可达90%以上。说明本系统是可行和实用的。

Research and Implementation of Automatic License Plate Recognition System

In order to alleviate the growing traffic pressure,improve the urban traffic management efficiency and enhance peoples＇ sense of security,this paper provides a method for license plate location based on texture and color information.By analyzing the horizontal projection of the statistical characteristics and features of the vertical projection to segment license plate characters,and by using adaptive and powerful learning ability BP neural network to recognize characters,an automatic plate recognition system is realized.The recognization time of plate in this system is less than 200 ms,and the precision of recognization is more than 90%.The experimental results verify the validity and practicality of this system.