一种噪声车牌的模糊预处理方法  

A Fuzzy Pretreatment Method for the Dirty Car License Plate with Noise

作　　者：魏运[1] 黄卫[2] 

机构地区：[1]北京城建设计研究总院,北京100037 [2]东南大学ITS研究中心,南京210096

出　　处：《交通运输工程与信息学报》2008年第3期85-90,共6页Journal of Transportation Engineering and Information

基　　金：江苏省自然科学基金项目(BK2004077);高等学校科技创新工程重大项目培育资金项目(705020)

摘　　要：本文提出了一种车牌模糊预处理方法,主要用于解决现场车牌图像模糊不清、对比度不强,以及传统二值化方法带来的噪声、粘联、变形等不理想现象。通过采用模糊增强技术来增强车牌图像的对比度,便于后续的分割、识别等操作;并提出应用模糊c均值算法来确定车牌图像二值化中的聚类阈值,从而实现对车牌图像的二值化,并将二值化结果与传统的Otsu二值化方法进行了对比。实验结果显示,应用本方法处理车牌噪声和粘联等情况具有较好的优越性。A method for fuzzy pretreatment of car license plate was presented in this paper. The main aim is to solve the faulty phenomenon of the spot car license plate, such as blur, weak contrast, noise, conglutination and hackle caused by the traditional binaryzation. The fuzzy enhancement was used to improve the contrast of license plate, which is convenient for the character segmentation and recognition. In addition, the fuzzy c-mean was presented to confirm the cluster threshold of the binaryzation of the car license plate,which could gained the image binaryzation. The method was contrasted with the traditional Otsu method, the experiment result showed that this method was more predominant than the traditional binaryzation method in reducing noise and distortion.

关 键 词：预处理 模糊增强 模糊C均值 噪声车牌 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]