欢迎来到天天文库
浏览记录
ID:36565945
大小:5.80 MB
页数:66页
时间:2019-05-12
《车牌字符识别关键技术研究及车牌识别系统实现》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库。
1、中山大学硕士学位论文车牌字符识别关键技术研究及车牌识别系统实现姓名:邓荣峰申请学位级别:硕士专业:交通信息工程及控制指导教师:李熙莹20090609车牌字符识别关键技术研究及车牌识别系统实现Title:LicensePlateCharacterRecognitionKeyTechniqueResearchandSystemImplementationMajor:TrafficInformationEngineeringandControlName:DengRongfengSupervisor:LiXiying(ass
2、oeiateprofessor)AbsttactLicenseplaterecognitionsystemisaveryimportantpartofintelligenttrafficsystem,itplaysallimportantroleinelectronicpoliceman,electronictollcollectionsystem,parkinglotmanagement,andregistrationofvehiclesincertainregions.Thispaperfocusesonstat
3、ionarysceneslikefreewaysandurbanroads,andproposessomepracticalsolutionstoskewcorrection,binarizationofimageandcharacterrecognition.AndthemaincontributionsofthispaperareaSfollows:Basedonthedetailanalysisofcurrentmethods,thispaperproposedanimprovementmethodofHoug
4、htransformtoenhancetheprecisionofobliquitydetection.“Reasonableobliquity'’isintroduced,andiftheobliquitydetectedbyHoughtransformisn’tareasonableone,thenthelinearfittingmethodisinvitedtodetecttheobliquity.ThispaperusesLoGoperatortogetthebinaryimageofthelicensepl
5、ate,andremovesthenoisewithregiongrowingmethod,avoidinglossoftheinformationundernon-uniformilluminationconditions,andrestrainingthenoisesbroughtbylocalbinarizationmethodfitswell.Afterreviewingcurrentfeatureextractionmethods,thispapertakes鲥dfeaturesandZemikemomen
6、tsascharacterfeatures.Gridfeaturescanreflectthedetailsoftheimage,whileZemikemomentscanreflecttheglobalinformationoftheimagesteadily.Withthesefeatures,minimumdistanceclassifierandBPNNareappliedtoclassifycharacters.Basedonthemainjobsmentionedabove,thispaperpropos
7、esanewLPRsystem,combinedwiththestudyoflicenseplatelocating,licensecolordistinguishing,andcharactersegmentation.Experimentsshowthatthealgorithmsinthispaperarereliableandthesystemisutilitarian.KeyWords:licenseplaterecognition;binarization;Zernikemoments;neuralnet
8、、j’rorkII中山大学硕士学位论文论文原创性声明内容本人郑重声明:所呈交的学位论文,是本人在导师的指导下,独立进行研究工作所取得的成果。除文中已经注明引用的内容外,本论文不包含任何其他个人或集体已经发表或撰写过的作品成果。对本文的研究作出重要贡献的个人和集体,均已在文中以明确方式标明。本人完全意识到本声明的法律结果由本人承担。学位论文作者签
此文档下载收益归作者所有