基于半监督学习的车牌识别方法研究 电子信息工程专业VIP免费

第1页课题名称：基于半监督学习的车牌识别方法研究摘要随着科技的发展，汽车作为交通工具也越来越普遍地被大众应用在日常生活当中。虽然我国的基础实施在不断完善，越来越多的高速公路，和大型停车场被建设，但其配套的车辆管理系统却相对落后，可以说人为的管理已经跟不上时代的潮流。为了解决上述问题，车牌自动识别技术应时而生，并发展成为现代道路系统的重要组成之一。它以计算机视觉处理、数字图像处理、模式识别等技术为基础，对摄像机所拍摄的车辆图像或者视频图像进行处理分析，进而得到每辆车的车牌号码和登记信息，可以实现盗抢车辆的管理、高速公路超速自动化的管理、电子警察、高速公路收费等等功能。这些对于维护交通安全和城市治安，实现城市交通智能化管理等都有着现实的意义。本文主要研究车牌识别的最后一道工序，对给定车牌照进行二值化处理后，然后将车牌进行取轮廓并采用垂直投影和模版匹配技术进行字符分割。将分割好的字符输入到字符识别模块。字符识别模块提取字符的相关特征，然后和模板库中的特征作比较，对车牌字符进行识别。最终得出车牌识别信息。关键字：灰度化二值化字符分割模板匹配垂直投影第2页ABSTRACTWiththedevelopmentofscienceandtechnology,automobilesasvehicleshavebecomemoreandmorewidelyusedbythegeneralpublicindailylife.Althoughourcountry’sbasicimplementationisconstantlyimproving,moreandmorehighwaysandlargeparkinglotsarebeingconstructed,thesupportingvehiclemanagementsystemsarerelativelybackward.Itcanbesaidthatman-mademanagementcannotkeepupwiththetrendofthetimes.Inordertosolvetheaboveproblems,licenseplateautomaticidentificationtechnologyshouldemergefromtimetotimeandbecomeoneoftheimportantcomponentsofthemodernroadsystem.Basedontechniquessuchascomputervisionprocessing,digitalimageprocessing,andpatternrecognition,itprocessesandanalyzesvehicleimagesorvideoimagescapturedbythecamera,andthenobtainslicenseplatenumbersandregistrationinformationforeachvehicle,andthenadoptssomefollow-upprocessingtechnologies.Itcanrealizethemanagementofstolenvehicles,themanagementofhighwayspeedingautomation,theelectronicpoliceandhighwaytollcollectionandotherfunctions.Thesehavepracticalsignificanceformaintainingtrafficsafety,urbansecurity,andintelligentmanagementofurbantraffic.Thispapermainlystudiesthelastprocessoflicenseplaterecognition.Afterbinarizationofagivenvehiclelicenseplate,thelicenseplateisprofiledandthecharactersegmentationisperformedusingverticalprojectionandtemplatematchingtechniques.Thesplitcharacterisinputtothecharacterrecognitionmodule.Thecharacterrecognitionmoduleextractstherelevantfeaturesofthecharacters,andthencomparesthemwiththefeaturesinthetemplatelibrarytoidentifythelicenseplatecharacters.Thefinallicenseplateidentificationinformation.KEYWORDS:Grayscale，Binarization，Charactersegmentation，Templatematching，Verticalprojection，第i页目录1绪论............................................................................................................11.1课题研究背景和意义.................................................................11.2车牌识别国内外发展现状.......................................................21.3本文研究内容...............................................................................31.4本文的组织结构..........................................................................32车牌的预处理.............................................................................................52.1颜色判断.........................................................................................52.1.1RGB颜色空间..............