一种新的基于字符形状特征的高效车牌识别算法  

High-performance license plate recognition method based on shape feature of characters

作　　者：胡金蓉[1,2] 周激流[1] 王玲[2] 郑秀清[2] 

机构地区：[1]四川大学计算机学院,成都610064 [2]四川师范大学计算机科学学院,成都610068

出　　处：《计算机应用研究》2010年第6期2398-2400,共3页Application Research of Computers

摘　　要：提出了一种基于字符的凹凸性、弯曲度和交点数等形状特征的车牌字符识别方法,它无须对字符作归一化、细化、倾斜校正这些预处理,也不需要样本图像,降低了字符畸变所产生的误识和拒识,减少了处理时间,提高了识别率。实验结果表明这是一种高效的车牌识别算法,是对现有车牌识别技术的有益补充。This paper proposed a simple and high-performance method of license plate recognition, which based on the shape feature of characters： convex and concave, bending and intersection. It didn’t need the normalization, thinning and orientation correction to the characters. Pre-processing to characters increased the rejection rate and error rate of license plate recognition because of the distortion resulted by the thinning and orientation correction. Experiments show that this method can improve the recognition rate and speed rapidly.

关 键 词：车牌识别 细化 形状特征 凹凸性 弯曲度 

分 类 号：TP391.43[自动化与计算机技术—计算机应用技术]