摘要
提出了一种基于NDRAM算法的车牌汉字识别方法.分别采用基于外积法和伪逆法的Hopfield网络和NDRAM算法对加有不同噪声水平的椒盐噪声的车牌汉字和具有不同程度缺失的车牌汉字进行了仿真识别.结果表明,NDRAM方法具有更高的识别率和更强的自适应性.
This paper proposes a method based on the NDRAM algorithm to recognize the Chinese characters on the license plates.After adding different levels of salt and pepper noise and different levels of partial loss on the Chinese characters,Hopfield neural network using Hebb rule,pseudoinverse method and NDRAM are applied to recognize respectively.Simulation results show that comparing with the other two methods,NDRAM exhibits higher recognition rate and greater adaptability.
出处
《东北师大学报（自然科学版）》
CAS
CSCD
北大核心
2015年第4期64-68,共5页
Journal of Northeast Normal University(Natural Science Edition)
基金
吉林省科技发展计划项目(20130103028JC)
关键词
NDRAM
联想记忆
模式识别
汉字
车牌
NDRAM
associative memory
pattern recognition
Chinese character
license plate
