摘要
提出了一种基于分级RBF神经网络的车牌字符识别方法,采用两级RBF神经网络结构,由一级网络识别后,根据识别结果和置信度,建立识别分布图,进行二级网络设计,确定了12个二级子网。RBF网络中自动确定隐层神经元数,无需实验调整。用大量样本对系统进行测试,车牌整体识别率达到了85.4%,通过对比性研究,验证了该方法的有效性和先进性。
A new recognition algorithm for license plate character based on multi-level RBF network is proposed.Two-level RBF network is adopted.According to recognition results from one-level network and the confidence levels,recognition distribution table is built,and two-level network is accordingly designed.As a result, 12 two-level sub-networks are formed.A large amount of samples are used for system test.Overall recognition accuracy is 85.4%.Through contrastive research,the method presented is proved to be effective and advanced.
出处
《计算机工程与应用》
CSCD
北大核心
2008年第30期213-216,共4页
Computer Engineering and Applications
基金
科技部国际合作重点项目(No.2003DF020009)。
关键词
车牌识别
径向基函数(RBF)网络
二级网络
识别率
license plate recognition
Radial Basis Function（RBF） network
two-level network
recognition accuracy
