摘要
对基于神经网络集成的汽车牌照识别的原理和方法进行了研究,并着重分析了现有技术的积极因素和潜在问题,提出了一种基于神经网络集成进行车牌文字识别的方法。在特征提取时采用了多种特征提取的方法,对提取的每种特征构建一个BP神经网络分别进行训练。最终待识别的字符将被神经网络集成进行识别。实践证明,利用该方法比单个神经网络识别有更高的识别率,具有较高的使用价值。
The research on the principle and method of the license plate recognition based on ensemble neural networks is carded on, the analysis of their positive effects and potential problems is emphasized on. A new method is proposed based on ensemble neural networks to recognize the license plate. We use combined approaches to extract the features. For every extracted feature, a corresponding BP neural network is set up and trained. At last, the image of symbols will be recognized by the ensemble neural networks. The experiments show that this method has a higher success rate and lower fail rate.
出处
《计算机工程与设计》
CSCD
北大核心
2007年第19期4741-4742,4746,共3页
Computer Engineering and Design
基金
广东技术师范学院2006年度青年基金项目(自然科学类)
关键词
BP神经网络
车牌识别
字符识别
神经网络集成
归一化
BP neural network
license plate recognition
character recognition
ensemble neural networks
normalization
