基于YCgCr颜色空间-肤色检测方法

YCgCr2009,30(19)44430Vezhnevets[1]()Gomez[2]H(HSV)-Y(RGB-Y)-Wr(Wr)PhungS.L[3]Shin[4]99Rein-LienHsu[5]YCgCr[7]1CgCrYCgCrYCbCr[8]YCgCrYCbCrYCbCr[9]YCgCrYcbCr(1)YCgCrYCbCryY[16,235]CgCr2008-10-162009-07-08(1980)(1963)[e-mail](450004)YCgCrY-CgY-CrYCbCrYCgCrYCgCrYCbCrYCgCr;;;;TP391.41A1000-7024(2009)19-4443-04AlgorithmofskindetectionbasedoncolorspaceYCgCrKANGJin-feng,WANGGuo-ying(InstituteofElectronicTechnology,PLAInformationEngineeringUniversity,Zhengzhou450004,China)AbstractAskincolordetectingalgorithmundervariantlightconditionispresented.Firstly,thefeatureofYCgCrcolorspacesystemisanalyzed,andaskincolorclassificationtechniqueisadoptedthatemploysBayesiandecisionwithcolorstatisticsdata,theskincolormodelisconstructedbytwolookuptablesinY-CgandY-Crsubspaces.AftercomparisonusingthesamemethoditsconcludedYCgCrcolorspaceisbetterthanYCbCrcolorspace.ExperimentalresultshavedemonstratedthatfalsedetectionrateandfalsedismissalrateatYCgCrcolorspacearebothlowerthanatYCbCrcolorspace,thismethodcanfitunderwiderlightcondition,anditsalsoeffectivetodetectskin-colorregionsfromthecolorimageswithcomplexbackground.KeywordsYCgCrColorspace;lookuptable(LUT);Bayesiandecision;skincolordetection;errorratesComputerEngineeringandDesign44442009,30(19)ComputerEngineeringandDesignCgCr[16,240]CgCbYCgCrRGBYCbCrRGB[7]=0128128+0.2990.1690.5000.5870.3310.4190.1140.5000.081(1)=0.00460.00460.004600.00150.00790.00630.0032016128128(2)=16128128+0.2560.3180.43920.50410.43920.36770.09790.12120.0714(3)=1.16431.16441.164700.94034.84151.596004.186116128128(4)(3)[0,1]RGBYCgCrYCgCrY[16,235],Cg,Cr[16,240](4)[0,1]RGB[10]2YCgCrYCgCr[11]Y-CgY-Cr2.1Y-CgY-CrY-CgY-CrY-CgY-Cr2.21=12(5)12121212[6]=21(6)(5)(6)={1,2}=(7)122[-1,1](5)0,+3Y-CgY-Cr[-1,1][10]12=3(8)i=1,2,,N21/2N2S==11,=11=121/2(9)3[0,1]=+12(10)2.3=0.700.0251(a)YCbVrCrCb250050100150200250010015020050250200150100500Y(b)YCgVrCrCg250050100150200250010015020050250200150100500YYCgCr2009,30(19)4445325006720010019013081802473Y-CgY-CrY-CgY-CrY[16,235],Cg[80,127],Cr[106,179](11)(12)(3)YCgCrYCbCrYCgCrY-CbY-CrY[16,235],Cb[89,127],Cr[106,183](12)(12)(1)333250020()22YCgCr=0.4549.7649%11.0781%YCbCr=0.47314.8724%12.9187%YCgCrYCbCr3(a)(d)3(b)(c)(e)(f)23(b)(c)(e)(f)(b)(e)YCbCr(c)(f)YCgCr(c)(f)(b)(e)YCgCr4YCgCrYY-CgY-CrYCgCr:[1]VezhnevetsV,SazonovV,AndreevaA.Asurveyonpixel-basedskincolordetectiontechniques[C].Moscow,Russia:ProcGrap-hicon-2003,2003:85-92.[2]GomezG.Onselectingcolorcomponentsforskindetection[C].Quebec,Canada:Proceedingsofthe16thInternationalConfe-renceonPatternRecognition,2002:961-964.[3]PhungSL,BouzerdoumA,ChaiD.Skinsegmentationusingcolorpixelclassification,analysisandcomparison[J].IEEETransactionsonPatternAnalysisandMachineIntelligence,2005,27(1):149-154.[4]ShinMC,ChangKI,TsapLV.Doescolorspacetransformationmakeanydifferenceonskindetection[C].IEEEWorkshoponApplicationsofComputerVision,2002.[5]Rein-LienHsu,MohamedAbdel-Mottaleb,AnilJainK.Facede-tectionincolorimage[C].IEEETransPatternAnalysisandMa-chineIntelligence,2002:696-706.YCgCr2YCbCrYCgCrYCbCr0.40.450.50.550.60.651009080706050403020100/%3(a)(b)(c)(d)(e)(f)44462009,30(19)ComputerEngineeringandDesign[6]DudaRO.[M].:,2003:1-15.[7]deDiOSJJ,GarciaN.FacedetectionbasedonanewcolorspaceYCgCr[C].ProceedingsofInternationa1ConferenceonImageProcessing,2003:909-912.[8]PhungSL,BouzerdoumA,ChaiD.AnovelskincolormodelinYCbCrcolorspaceanditsapplicationtohumanfacedetection[C].ProceedingsofInternationalConferenceonImageProce-ssing,2002:289-292.[9],,.YCgCr[J].,2007,34(5):228-233.[10],.YCbCr[J].,2006,11(1):47-52.[11],,.[J].,2002,38(24):123-125.31,1(2,2)4Petri1,1Messageproce-ssingunit(2,2)ref4(Messageprocessingunit5)Petri5(1,1)Messageprocessingunit(2,2)(4,4')[1](1)(11)1(1,1)(2,2)(1,1)Messageprocessingunit(2,2)4PetriBDIBDI:[1].Petri[M].:,2005:1-103.[2].[M].:,2000:1-90.[3].[M].:,2002:24-138.[4],.[J].,2005,28(1):1-7.[5].[D].:,2001.[6],,,.[J].,2001,24(5):529-535.[7],,.Petri[J].,2003,15(s):114-118.[8].PetriNetsBDIAgent[J].,2005,31(3):23-25.[9]YuZhenhua,LiZhiwu.Softwarearchitecturedescriptionlan-guagebasedonobject-orientedpetrinets[J].JournalofXi'anJiaotongUniversity,2004,38(12):1236-1239.[10]Jimhez-OchoaI,BeEovichO,Ramirez-TreviiioA,etal.Imple-mentingBDIagentsusingpetrinets[C].ProceedingsoftheIEEEInternationalConferenceonSystems,ManandCybernetics,2003:286-291.[11]OsamuKatai,KentaroToda,HiroshiKawakami.Decentralizedcontrolofmulti-agentsystemsbasedonmodallogicsandex-tendedhigherorderpetrinets[C].Springer-VerlagBerlinHeidel-berg2001,LNCS2252,2001:182-190.[12]MichaelKohler,DanielMoldt,HeikoRolke.Modellingthestruc-tureandbehaviourofpetrinetagents[C].ColomJM,KoutnyM.Springer-VerlagBerlinHeidelberg,2001:224-241.[13]DaltonSereyGuerrero,JorgeCAdeFigueiredo,AngeloPer-kusich.Anobject-basedmodularCPNapproach:itsapplicationtothespecificationofacooperativeeditingenvironment[C].Springer-VerlagBerlinHeidelberg.ConcurrentOOPandPN,2001:338-354.[14]CaiLinqin,MeiTao,SunYining,etal.Modelingandanalyzingmulti-agenttaskplansforintelligentvirtualtrainingsystemusingpetrinets[C].ProceedingsoftheWorld