資源描述:
《圖像五值化和基于三值圖像車(chē)牌識(shí)別》由會(huì)員上傳分享,免費(fèi)在線(xiàn)閱讀,更多相關(guān)內(nèi)容在工程資料-天天文庫(kù)。
1、JournalofSoftware軟件學(xué)報(bào)?圖像五值化與基于三值圖像的車(chē)牌識(shí)別張忠義1+1(??谟蘩袝?huì)教育科技有限公司,海南海口570203)Image5-Value-SegmentationandLicensePlateRecognitionbasedon3-valueimage*ZHANGZhong-Yi1+1(HaikouYulaohuiEducationTechnologyCo.,Ltd,Haikou570203,China)+Correspondingauthor:Phn:+86-898-36393679,Fax:+8
2、6-898-36393679,E-mail:yulaohui@qq.comReceived;AcceptedAbstract:Image5-value-segmentationisaimagesimplificationmethodusedtomapRGBspaceintoYBspace,bytheformulationY=(R+G)/2andthenconvertcolorinYBspacetofivevalues:black、blue、gray、yellowandwhite.ForChineselicenseplaterec
3、ognition,fivevaluescanfurtherbecompressedtothreevaluesbycombiningblackandblue,whiteandyellowtogetherseparately.Themethodproposedhastwoapplicationscenarios:characterrecognitionbasedon3-valueimageandlicenseplatelocationbyinversingsegmentationinformationbasedon3-valueim
4、age.Inthefirstscenario,Thereservedgrayin3-valueimagecanbeclassifiedintoeithertheplatebackgroundcolorortheforegroundcharactercolorforthecharacterrecognitionaccordingtothelicenseplatebackgroundcolorrecognized.Itcanmakeupfortheinformationlossintheprocessofimagebinarizat
5、ionbydeterminingthreshold,whichisalwaysinfluencedbytheunevenilluminationordirtyspots.Doingthishelpstoimprovetheaccuracyofcharacterrecognition.Inthesecondscenario,candidatelocationoflicenseplatecanbefoundinthepremiseoftraversingtheimageefficientlyifcharactersegmentati
6、oncanbeperformedbasedonprojectionsofthe3-valueimage.Thismethodcandetectcharacterspacingfirstlytomeettheneedsofusers’andeliminatefalselicenseplate,makesuretheaccuracyoflicenseplatelocationwhichdoesn’tdependonlicenseplatecharacter;itcanalsoreducethetruenegativebyignori
7、ngtheinferencebroughtbyplatefeatures.Keywords:5-valuesegmentation;licenseplaterecognition摘要:圖像五值化是通過(guò)公式Y(jié)=(R+G)/2,將RGB空間映射到Y(jié)B空間,再將YB空間簡(jiǎn)化成黑、藍(lán)、灰、黃、白五種值的圖像簡(jiǎn)化方法。針對(duì)中國(guó)車(chē)牌,通過(guò)黑、藍(lán)合并,白、黃合并,圖像可進(jìn)一步簡(jiǎn)化成三值?;谌祱D像的車(chē)牌識(shí)別包括基于三值圖像的字符識(shí)別和基于三值圖像按分割反推的車(chē)牌定位。三值圖像保留的中間灰色可根據(jù)識(shí)別的車(chē)牌底色視同字符識(shí)別所需的車(chē)牌底色或字符
8、色,從而能彌補(bǔ)受光照不均勻或車(chē)牌“臟”影響按閾值選取進(jìn)行圖像二值化可能導(dǎo)致的圖像信息缺失,有利于提高字符識(shí)別的正確率?;谌祱D像按分割反推的車(chē)牌定位是指在確保圖像遍歷高效的前提下,通過(guò)基于三值圖像的投影如果可以進(jìn)行字符分割則得到一個(gè)候選車(chē)牌定位。其好處是可以先