复杂环境中的车牌字母和数字识别研究  

Research on Recognition of Letters and Figures on Vehicles' License Plates under Complex Condition

作　　者：王中华[1] 陈三宝[1] 周开军[1] 

机构地区：[1]武汉理工大学自动化学院

出　　处：《微型电脑应用》2006年第3期7-8,25,共3页Microcomputer Applications

摘　　要：当前车牌识别的研究大都是针对正常环境进行的,对于复杂环境下的车牌难以达到识别要求。本文提出一种用BP神经网络构造并行神经网络的车牌字母和数字识别方法,利用PVM网络在虚拟并行平台上实现了并行神经网络,最后对复杂现场环境下获取的车牌进行了实验。实验结果证明,该算法具有良好的性能,能在28ms内实时准确的识别车牌字母和数字。The current research on vehicles' license plates(LP)recognition is almost processed under normal condition. However, it is difficult to meet the need of LP recognition under complex condition. In this paper a method of LP recognition based on parallel neural network is presented, which is constructed by the standard BP neural network. Based on PVM network. the parallel neural network is implemented on the virtual parallel platform, and the license plates got under complex condition are experimented in the end. Experimental results prove that this method has nice performance-being able to recognize the letter and figure of LP correctly in 28ms.

关 键 词：牌照识别 BP神经网络 并行神经网络 PVM网络 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]