摘要
为了进行车牌字符的有效识别,提出了一种分级径向基函数神经网络的车牌识别算法,识别网络由两级构成,一级径向基神经网络用于汉字、字母、混合和数字的粗分类;二级子网用于对字母网络、混合网络和数字网络内部易混字符再进行精确识别。实验结果表明,提出的方法有效地提高了识别的精度,而且平均运行时间减少。
In order to carry out license plate identification effectively, a new method of license plate identification based on a two-level neural network is presented. The first-level network is used to classify mixed Chinese characters, letters, and numbers roughly, and the secondary sub-network is used to do exact recognition of confusable characters in mixed network again. The experimental results show that the proposed new method effectively improved the accuracy of license plate identification, and reduced an average time of license plate identification.
出处
《现代电子技术》
2011年第1期207-210,共4页
Modern Electronics Technique
基金
国家高技术研究发展计划(863计划)第四批课题(2008AA11A134)
四川省教育厅基础应用研究课题(2009ZX002)基金的部分资助
关键词
车牌识别
径向基函数神经网络
二级网络
识别率
license plate identification
RBF network
two-level network
recognition accuracy