参考文献01-A License Plate-Recognition Algorithm for Intelligent Transportation System Applications(1)

《参考文献01-A License Plate-Recognition Algorithm for Intelligent Transportation System Applications(1)》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、IEEETRANSACTIONSONINTELLIGENTTRANSPORTATIONSYSTEMS,VOL.7,NO.3,SEPTEMBER2006377ALicensePlate-RecognitionAlgorithmforIntelligentTransportationSystemApplicationsChristosNikolaosE.Anagnostopoulos,Member,IEEE,IoannisE.Anagnostopoulos,Member,IEEE,VassiliLoumos,Memb

2、er,IEEE,andEleftheriosKayafas,Member,IEEEAbstractInthispaper,anewalgorithmforvehiclelicensearterialmanagementsystemsfortrafficsurveillance.Moreover,plateidentificationisproposed,onthebasisofanoveladaptiveasincreasedsecurityawarenesshasmadetheneedforvehicle-imag

3、esegmentationtechnique(slidingconcentricwindows)andbasedauthenticationtechnologiesextremelysignificant,theconnectedcomponentanalysisinconjunctionwithacharacterproposedsystemmaybeemployedasaccesscontrolsystemrecognitionneuralnetwork.Thealgorithmwastestedwith133

4、4natural-scenegray-levelvehicleimagesofdifferentbackgroundsformonitoringofunauthorizedvehiclesenteringprivateareas.andambientillumination.Thecamerafocusedintheplate,whileThelicenseplateremainsastheprincipalvehicleidentifiertheangleofviewandthedistancefromtheve

5、hiclevariedac-despitethefactthatitcanbedeliberatelyalteredinfraudsitua-cordingtotheexperimentalsetup.Thelicenseplatesproperlytionsorreplaced(e.g.,withastolenplate).Therefore,ITSsrelysegmentedwere1287over1334inputimages(96.5%).TheopticalheavilyonrobustLPRsyste

6、ms.Thefocusofthispaperisoncharacterrecognitionsystemisatwo-layerprobabilisticneuralnetwork(PNN)withtopology108-180-36,whoseperformancefortheintegrationofanovelsegmentationtechniqueimplementedentireplaterecognitionreached89.1%.ThePNNistrainedtoinanLPRsystemabl

7、etocopewithoutdoorconditionsifpara-identifyalphanumericcharactersfromcarlicenseplatesbasedmeterizedproperly.Specifically,thenovelcontributionsareondataobtainedfromalgorithmicimageprocessing.Combiningtheabovetworates,theoverallrateofsuccessforthelicense-1)anove

8、lsegmentationtechniquenamedslidingconcentricplate-recognitionalgorithmis86.0%.Areviewintherelatedwindows(SCWs)usedforfasterdetectionofregionsofliteraturepresentedinthispaperrevealsthatbet