车牌识别算法的研究与实现  

Research and Realization of License Plate Recognition Arithmetic

作　　者：何铁军[1] 张宁[1] 黄卫[1] 

机构地区：[1]东南大学智能运输系统研究中心,江苏南京210096

出　　处：《公路交通科技》2006年第8期147-149,159,共4页Journal of Highway and Transportation Research and Development

基　　金：江苏省自然科学基金资助项目(BK2004077)

摘　　要：介绍一种基于KL变换的车牌识别算法。算法主要由车牌分割和车牌字符识别组成。车牌分割算法利用车牌的纹理特征和形状特征来定位车牌,并采用投影的方法和先验知识实现车牌的倾斜校正和字符切分;车牌字符识别算法采用一种最优变换,KL变换来提取字符图像的特征值和特征向量,并通过计算待识别字符与各样本字符特征值的欧氏距离来实现对该字符的分类。试验结果表明此方法是正确的。The authors present vehicle license plate recognition algorithm based on KL transform .This algorithm is made up of two parts： plate segmentation and character recognition.In the license plate segmentation algorithm, the character of texture and shape of the plate is utilized to locate the license plate, and image projection and prior knowledge is utilized to detect the skew of the license plate and extract characters from it. In the character recognition algorithm, KL transform, a kind of optimal transform, is applied to extract the eigenvalue and eigenvector from character images. Euler distances of this eigenvalue between recognizing character and sample characters are computed and compared to determine the classification. The validity of this algorithm is proved by test..

关 键 词：车牌识别 车牌定位 KL变换 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]