0,、、、,。,,、,。、,。,,,。,、。,,。,,,,。,。;;;TP391.4A1000-7857(2010)20-0112-061,1,21.,1500272.,150001AdvancesinResearchonEdgeDetectionofImages:2010-01-25;:2010-09-12:(GZ08A109);(11541323):,,,:g-cy@163.com;(),,,:ztf991@163.comGAOChaoyang1,ZHANGTaifa1,QUYanan21.SchoolofScience,HeilongjiangInstituteofScienceandTechnology,Harbin150027,China2.HarbinRadioandTVUniversity,Harbin150001,ChinaAbstractTheimageedgeisoneofthemostfundamentalandsignificantfeaturesoftheimage.Theedgecontainsagreatdealofusefulinformationonimportantcontoursandlocationsandofferspeopleimportantparameterstodescribeandrecognizeobjectsintheimage.Edgedetectionisoneofthemostbasicoperationsinimageprocessing,imageanalysis,andcomputervision.Itisoneofthebasicmethodsforpatternrecognitionandimageinformationextraction.Researchonedgedetectionhasalonghistory,andtherehavebeenmanyalgorithmsproposedforit.However,allofthemethodsforedgedetectionstillhavesomedrawbacksandcannotdetecttheoptimaledgesinsomecases.Itisdifficulttodevelopageneralmethodforedgedetection.Edgedetectionisastudyhotspotinthefieldofimageprocessing.Newfeasibletheoryandmethodsareproposedeveryday.Themainresearchmethodsforthestudyofedgedetectionaretomakeimprovementstoexistingmethodsortofindnewmethodsforedgedetectionwithspecialapplicationrequirements.Thispaperillustratesthesignificanceoftheedgedetectionandsummarizestheelementarystepsofedgedetectionandthegeneralrequirementsfortheobtainededge.Thenitreviewsthevariousmethodsofedgedetection,analyzesthemeritsandthedeficienciesofthosemethods,andsummarizestherecentdevelopmentofedgedetection.Finally,itanalyzesallkindsofcommonedgeevaluationmethodsavailableatpresently.Thispapermayserveasavaluablereferenceforfuturestudiesandapplicationsofedgedetection.Keywordsedgedetection;relaxativeiteration;deformablemodel;mathematicalmorphology(Reviews)1122010,28(20)[1]。,,、、、,。。1959,Julez[2];1965,Roberts[3]。50,,,,。,,,,。1,[4],。,,,。4,1。①。,,,。②。,。③。,。④。,,。。p(x,y),(1)[5]:p(x,y)=0.3R+0.59G+0.11B(1),R、G、B、、。,[6]:①;②;③;④;⑤。,,,。2。。,,,、、、,,。2.1,,。(),(),()。2.1.1,、Roberts、PrewittSobel,LaplaceWallis[7]。f(x,y),:grad(x,y)=坠f坠x坠f坠y T=[GxGy]Tgrad(x,y)=G2x+G2y姨,θ=arctanGxGy姨姨。,:grad(x,y)=max(f′x,f′y)grad(x,y)=f′x+f′y,f′x=f(x+1,y)-f(x,y),f′y=f(x,y+1)-f(x,y)。,。2(a),,。,,。1965,RobertsRoberts,,2(c),。,1970,Prewitt,2×23×3,2(d)。SobelPrewitt,4,2(e),(Reviews)2010,28(20)113,,。Laplace,,(2):Δ2f(x,y)=坠2f(x,y)坠x2+坠2f(x,y)坠y2(2),(2)Δ2f(x,y)=f(x+1,y)+f(x-1,y)+f(x,y+1)+f(x,y-1)-4f(x,y)2(b),,,。WallisLaplace,。2.1.2,。。1980,Marr[8]LaplaceLoG()。,LoG,,M(x,y)=Δ2(G(x,y)茚f(x,y))=(Δ2G(x,y))茚f(x,y)(3),Δ2G(x,y)=1πδ4x2+y22δ2-221exp-x2+y22δ222,,,,3[9]。LOG,。,,[10]。1986,Canny[11],,,。Canny,Canny:①;②;③,,;④,T1T2(T1T2),T1,T2,。CannyRoberts、Prewitt、SobelKirsch,,,。CannyDeriche,。2.1.3,。,,。,,,[12]。。,,,;,,。2080,X。2.22.2.1,,。,。1992,Lawrence[11]Fourier,Bayes,,。,。1970,Prewitt,n,n12,[13]ε=x,y∈ΔsΣ[p(x,y)-f(x,y)]2(4)(Reviews)1142010,28(20),p(x,y),f(x,y),Δs,mn×n,。2.2.2,,、,,,,4。,,、、。、,,,,,,。2.2.3,,,,,Rosenfeld,Marr,Within,。Mallat[14]1992Canny,Fourier,,,,,。Mallat,,:①;②,;③,,;④;⑤。2.2.42090,,,,[15]。。snake,,,;,。,。,。,,,[16-17]。,。OsherSethian1988[18],。2.2.5,,,,。,、、、,。5[19]。,,BP。BP,BP,,,,。2.2.6。,、。2080,PalKing[20],,。6。(Reviews)2010,28(20)1152090,Cheng[21],C。J.K.Udupa[22],,,,,。2.2.7、、,。,[23]。,。A,,6:①:MEO1=fA-f;②:MEO2=f-fΘA;③:MEO3=fA-fΘA;④:MEO4=f-f°A;⑤:MEO5=f·A-f;⑥:MEO6=f·A-f°A。,A。,,,,。2.2.8,。[24],。[25],,。[26],,,。[27],,。[28]CT,,。[29],。[30],,,,。[31],,Sobel。[32],,,。3。1981,Kitchen[33],,、,。1992,Venkatesh[34]、、,Kitchen,,。1998,[35-36],,。、,。2003,[37],A、4B、8CC/A、C/B,,。,,、。4,,,。,,,,。,:①;②、;③;④;⑤。(References)[1],.[J].,2006,42(30):88-91.WeiWeibo,RuiXiaoting.ComputerEngineeringandApplications,2006,42(30):88-91.[2]JulezB.AmethodofcodingTVsignalsbasedonedgedetection[J].BellSystemTech,1959,38(4):1001-1020.[3]RobertsLD.Machineperceptionofthree-dimensionsolidsinopticalandelectro-optimalinformationprocessing[C].Cambridge,Massachusetts:MassachusettsInstituteofTechnologyPress,1966:157-197.[4]HildrethEC.Edgedetection[M].Cambridge,Massachusetts:MassachusettsInstituteofTechnology,ArtificialIntelligenceLaboratory,1985.[5],.[J].,2005,35(12):980-982.LiuQitao,ZhangJun.Laser&Infrared,2005,35(12):980-982.[6],,.[J].,2005,31(3):415-419.(Reviews)1162010,28(20)DuanRuiling,LiQingxiang,LiYuhe.OpticalTechnique,2005,31(3):415-419.[7],,,.[J].,2007(8):23-26.RenYibin,WangZiyan,YuJihong,etal.Computer&InformationTechnology,2007(8):23-26.[8]MarrD,HildrethE.Theoryofedgedetection[C]//ProceedingoftheRoyalSocietyofLondon.SeriesB,BiologicalSciences.1980,207(1167):187-217.[9],,.Log[J].,2008,44(22):183-185.ZhangTaifa,ChengDongxu,ShiDuanyin.ComputerEngineeringandApplications,2008,44(22):183-185.[10]ClarkJJ.Authenticatingedgeproducedbyzero-crossingalgorithm[J].IEEETransactiononPatternAnalysisandMachineIntelligence,1998,11(1):43-57.[11]CannyJ.Acomputationalapproachtoedgedetection[J].IEEETransactiononPatternAnalysisandMachineIntelligence,1986,8(6):679-698.[12],.[J].,2002,15(2):192-204.LinYao,TianJie.PatternRecognitionandArtificialIntelligence,2002,15(2):192-204.[13].CT[D].:,2008.KangYi.Studyingofedgedetectionofmedic