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.
出处
《计算机应用研究》
CSCD
北大核心
2010年第6期2398-2400,共3页
Application Research of Computers
关键词
车牌识别
细化
形状特征
凹凸性
弯曲度
license plate recognition
thinning
shape feature
convex and concave feature
bending
