电子信息与通信工程专业英语课件chapter9

电子信息与通信工程专业英语Chapter9DigitalImageProcessing

Interestindigitalimageprocessingmethodsstemsfromtwoprincipalapplicationareas:improvementofpictorialinformationforhumaninterpretation;andprocessingofimagedataforstorage,transmission,andrepresentationforau­tonomousmachineperception.Thischapterhasseveralobjectives:(1)todefinethescopeofthefieldthatwecallimageprocessing;(2)togiveahistoricalper­spectiveoftheoriginsofthisfield;(3)todiscussbrieflytheprincipalapproachesusedindigitalimagepro­cessing;(4)togiveanoverviewofthecomponentscontainedinatypical,general-purposeimageprocessingsystem.9.1WhatIsDigitalImageProcessing1Animagemaybedefinedasatwo-dimensionalfunction,,wherexandyarespatial(plane)coordinates,andtheamplitudeoffatanypairofcoordinatesiscalledtheintensityorgrayleveloftheimageatthatpoint.Whenx,y,andtheamplitudevaluesoffareallfinite,discretequantities,wecalltheimageadigitalimage.Thefieldofdigitalimageprocessingreferstoprocessingdigitalimagesbymeansofadigitalcomputer.Notethatadigitalimageiscomposedof

2afinitenumberofelements,eachofwhichhasaparticularlocationandvalue.Theseelementsarereferredtoaspictureelements,imageelements,pels,andpixels.Pixelisthetermmostwidelyusedtodenotetheelementsofadigitalimage.Visionisthemostadvancedoneinoursenses,soitisnotsurprising3thatimagesplaythesinglemostimportantroleinhumanperception.However,unlikehumans,whoarelimitedtothevisualbandoftheelectromagnetic(EM)spectrum,imagingmachinescoveralmosttheentireEMspectrum,rangingfromgammatoradiowaves.Theycanoperateonimagesgeneratedbysourcesthathumansarenotaccustomedtoassociatingwithimages.Theseincludeultrasound,electronmicroscopy,andcomputer-generatedimages.Thus,digitalimageprocessingpassesawideandvariedfieldofapplications.Thereisnogeneralagreementamongauthorsregardingwhereimageprocessingstopsandotherrelatedareas,suchasimageanalysisandcomputervision,start.Sometimesadistinctionismadebydefiningimageprocessingasadisciplineinwhichboththeinputandoutputofaprocessareimages.Webelievethistobealimitingandsomewhatartificialboundary.Forexample,underthisdefinition,eventhetrivialtaskofcomputingtheaverageintensityofanimage(whichyieldsasinglenumber)wouldnotbeconsideredasanimageprocessingoperation.Ontheotherhand,therearefieldssuchascomputervisionwhoseultimategoalistousecomputerstoemulatehumanvision,includinglearningandbeingabletomakeinferencesandtakeactionsbasedonvisualinputs4.Thisareaitselfisabranchofartificialintelligence(AI)whoseobjectiveistoemulatehumanintelligence.ThefieldofAIisinitsearlieststagesofinfancyintermsofdevelopment,withprogresshavingbeenmuchslowerthanoriginallyanticipated.Theareaofimageanalysis(alsocalledimageunderstanding)isinbetweenimageprocessingandcomputervision.Therearenoclear-cutboundariesinthecontinuumfromimageprocessingatoneendtocomputervisionattheother.However,oneusefulparadigmistoconsiderthreetypesofcomputerizedprocessesinthiscontinuum:low-,mid-,andhigh-levelprocesses.Low-levelprocessesinvolveprimitiveoperationssuchasimagepreprocessingtoreducenoise,contrastenhancement,andimagesharpening.Alow-levelprocessischaracterizedbythefactthatbothitsinputsandoutputsareimages.Mid-levelprocessingonimagesinvolvestaskssuchassegmentation(partitioninganimageintoregionsorobjects),descriptionofthoseobjectstoreducethemtoaformsuitableforcomputerprocessing,andclassification(recognition)ofindividualobjects.Amid-levelprocessischaracterizedbythefactthatitsinputsgenerallyareimages,butitsoutputsareattributesextractedfromthoseimages(e.g.,edges,contours,andtheidentityofindividualobjects).Finally,higher-levelprocessinginvolves"makingsense"ofanensembleofrecognizedobjects,asinimageanalysis,and,atthefarendofthecontinuum,performingthecognitivefunctionsnormallyassociatedwithvision.Basedontheprecedingcomments5,weseethatalogicalplaceofoverlapbetweenimageprocessingandimageanalysisistheareaofrecognitionofindividualregionsorobjectsinanimage.Thus,whatwecallinthischapterdigitalimageprocessingpassesprocesseswhoseinputsandoutputsareimagesand,inaddition,passesprocessesthatextractattributesfromimages,uptoandincludingtherecognitionofindividualobjects.Asasimpleillustrationtoclarifytheseconcepts,considertheareaofautomatedanalysisoftext.Theprocessesofacquiringanimageoftheareacontainingthetext,preprocessingthatimage,extracting(segmenting)theindividualcharacters,describingthecharactersinaformsuitableforcomputerprocessing,andrecognizingthoseindividualcharactersareinthescopeofwhatwecalldigitalimageprocessinginthischapter.Makingsenseofthecontentofthepagemaybeviewedasbeinginthedomainofimageanalysisandevencomputervision,dependingonthelevelofcomplexityimpliedbythestatement"makingsense".Aswilleevidentshortly,digitalimageprocessing,aswehavedefinedit,isusedsuccessfullyinabroadrangeofareasofexceptionalsocialandeconomicvalue.Theconceptsdevelopedinthefollowingsectionsarethefoundationforthemethodsusedinthoseapplicationareas.Newwordsandphrasesimage[ˈimidʒ]n.vt.图像；想象；肖像；偶像作……的图像；想象；反映；象征stemfrom起源于perception[pəˈsepʃ(ə)n]n.知觉；【生理】感觉；看法；洞察力；获取imageprocessing【计】【遥感】图像处理；图像加工perspective[pəˈspektiv]n.adj.视角，观点；远景；透视图透视的spatial[ˈspeiʃ(ə)l]adj.空间的；存在于空间的；受空间条件限制的amplitude[ˈæmpliˌtju:d]n.振幅；丰富，充足；广阔coordinate[kəuˈɔ:dineit]n.v.【数】坐标；相配的衣物使协调；使调和pelsn.像素；图元pixel[ˈpiksəl]n.像素；像素点(pixel的复数)electromagnetic[iˌlektrə(ʊ)mægˈnetik]adj.电磁的spectrum[ˈspektrəm]n.光谱；频谱；范围；余象ultrasound[ˌʌltrəsaʊnd]n.超声；超音波electronmicroscopy【电子】【光】电子显微镜学，电子显微镜discipline[ˈdisiplin]n.学科；纪律；训练；惩罚artificial[ɑ:tiˈfiʃ(ə)l]adj.人造的；仿造的；虚伪的；非原产地的；武断的intensity[inˈtensiti]n.强度；强烈【电子】亮度；紧张emulate[ˈemjuleit]vt.n.仿真；模仿；尽力赶上；同……竞争仿真；仿效inference[ˈinf(ə)r(ə)ns]n.推理；推论；推断visual[ˈviʒjuəl;-zj-]adj.视觉的，视力的；栩栩如生的artificialintelligence(AI)人工智能infancy[ˈinf(ə)nsi:]n.初期；婴儿期；幼年anticipated[ænˈtisipeitid]adj.预期的；期望的，预料(anticipate的过去分词)；盼望paradigm[ˈpærədaim]n.范例；词形变化表contrastenhancement【电子】对比增强；增衬器；对比强化imagesharpening图像锐化；图像清晰化classification[ˌklæsifiˈkeiʃ(ə)n]n.分类；类别，等级finite[ˈfainait]adj.n.有限的；限定的有限之物discrete[disˈkri:t]adj.n.离散的，不连续的分立元件；独立部件Notes1.本节主要介绍数字图像处理的概念。2.

becomposedof由……组成，由……构成。此句可翻译为：值得提及的是数字图像是由有限的元素组成的，每一个元素都有一个特定的位置和幅值。3.itisnotsurprisingthat…显而易见的是，毫无疑问。此句可翻译为：视觉是人类最高级的感知器官，所以，毫无疑问图像在人类感知中扮演着最重要的角色。4.此句为一简单句，句型是there+be句型，whose引导的定语从句修饰fields。此句可翻译为：另一方面，有些领域(如计算机视觉)研究的最高目标是用计算机去模拟人类视觉，包括理解和推理并根据视觉输入采取行动等。5.

basedontheprecedingcomments根据上述讨论，根据上述分析。此句可翻译为：根据上述讨论可知，图像处理和图像分析两个领域合乎逻辑的重叠部分是图像中特定区域或物体的识别这块。9.2TheOriginsofDigitalImageProcessing1Oneofthefirstapplicationsofdigitalimageswasinthenewspaperindustry,whenpictureswerefirstsentbysubmarinecablebetweenLondonandNewYork.IntroductionoftheBartlanecablepicturetransmissionsystemintheearly1920sreducedthetimerequiredtotransportapictureacrosstheAtlanticfrommorethanaweektolessthanthreehours.Specializedprintingequipmentcodedpicturesforcabletransmissionandthenreconstructedthematthereceivingend.Someoftheinitialproblemsinimprovingthevisualqualityoftheseearlydigitalpictureswererelatedtotheselectionofprintingproceduresandthedistributionofintensitylevels.Theprintingmethodwasabandonedtowardtheendof1921infavorofatechniquebasedonphotographicreproductionmadefromtapesperforatedatthetelegraphreceivingterminal.Theimprovementsovertheprintingmethodareevident,bothintonalqualityandinresolution.TheearlyBartlanesystemswerecapableofcodingimagesinfivedistinctlevelsofgray.Thiscapabilitywasincreasedto15levelsin1929.Duringthisperiod,introductionofasystemfordevelopingafilmplatevialightbeamsthatweremodulatedbythecodedpicturetapeimprovedthereproductionprocessconsiderably.Althoughtheexamplesjustcitedinvolvedigitalimages,theyarenotconsideredasdigitalimageprocessingresultsinthecontextofourdefinitionbecausecomputerswerenotinvolvedintheircreation.Thus,thehistoryofdigitalimageprocessingisintimatelytiedto2thedevelopmentofthedigitalcomputer.Infact,digitalimagesrequiresomuchstorageandcomputationalpowerthatprogressinthefieldofdigitalimageprocessinghasbeendependentonthedevelopmentofdigitalcomputersandofsupportingtechnologiesthatincludedatastorage,display,andtransmission.TheideaofacomputergoesbacktotheinventionoftheabacusinAsiaMinor,morethan5000yearsago.Morerecently,thereweredevelopmentsinthepasttwocenturiesthatarethefoundationofwhatwecallacomputertoday3.However,thebasisforwhatwecallamoderndigitalcomputerdatesbacktoonlythe1940swiththeintroductionbyJohnvonNeumannoftwokeyconcepts:(1)amemorytoholdastoredprogramanddata,and(2)conditionalbranching4.Thesetwoideasarethefoundationofacentralprocessingunit(CPU),whichisattheheartofcomputerstoday.StartingwithvonNeumann,therewereaseriesofkeyadvancesthatledtocomputerspowerfulenoughtobeusedfordigitalimageprocessing.Briefly,theseadvancesmaybesummarizedasfollows:(1)theinventionofthetransistorbyBellLaboratoriesin1948;(2)thedevelopmentinthe1950sand1960softhehigh-levelprogramminglanguagesCOBOL(CommonBusiness-OrientedLanguage)andFORTRAN(FormulaTranslator);(3)theinventionoftheintegratedcircuit(IC)atTexasInstrumentsin1958;(4)thedevelopmentofoperatingsystemsintheearly1960s;(5)thedevelopmentofthemicroprocessor(asinglechipconsistingofthecentralprocessingunit,memory,andinputandoutputcontrols)byIntelintheearly1970s;(6)introductionbyIBMofthepersonalcomputerin1981;and(7)progressiveminiaturizationofcomponents,startingwithlargescaleintegration(LI)inthelate1970s,thenverylargescaleintegration(VLSI)inthe1980s,tothepresentuseofultralargescaleintegration(ULSI).Concurrentwiththeseadvancesweredevelopmentsintheareasofmassstorageanddisplaysystems,bothofwhicharefundamentalrequirementsfordigitalimageprocessing.Thefirstcomputerspowerfulenoughtocarryoutmeaningfulimageprocessingtasksappearedintheearly1960s.Thebirthofwhatwecalldigitalimageprocessingtodaycanbetracedtotheavailabilityofthosemachinesandtheonsetofthespaceprogramduringthatperiod.Ittookthecombinationofthosetwodevelopmentstobringintofocusthepotentialofdigitalimageprocessingconcepts.WorkonusingcomputertechniquesforimprovingimagesfromaspaceprobebeganattheJetPropulsionLaboratory(Pasadena,California)in1964whenpicturesofthemoontransmittedbyRanger7wereprocessedbyacomputertocorrectvarioustypesofimagedistortioninherentintheon-boardtelevisioncamera.Figure9.1showsthefirstimageofthemoontakenbyRanger7onJuly31,1964at9:09a.m.EasternDaylightTime(EDT),about17minutesbeforeimpactingthelunarsurface(themarkers,calledreseaumarks,areusedforgeometriccorrections).ThisalsoisthefirstimageofthemoontakenbyaU.S.spacecraft.TheimaginglessonslearnedwithRanger7servedasthebasisforimprovedmethodsusedtoenhanceandrestoreimagesfromtheSurveyormissionstothemoon,theMarinerseriesofflybymissionstoMars,theApollomannedflightstothemoon,andothers.Figure9.1ThefirstpictureofthemoonbyaU.S.spacecraftInparallelwith5spaceapplications,digitalimageprocessingtechniquesbeganinthelate1960sandearly1970stobeusedinmedicalimaging,remoteearthresourcesobservations,andastronomy.Theinventionintheearly1970sofcomputerizedaxialtomography(CAT),alsocalledcomputerizedtomography(CT)forshort,isoneofthemostimportanteventsintheapplicationofimageprocessinginmedicaldiagnosis.Computerizedaxialtomographyisaprocessinwhicharingofdetectorsencirclesanobject(orpatient)andanX-raysource,concentricwiththedetectorring,rotatesabouttheobject6.TheX-rayspassthroughtheobjectandarecollectedattheoppositeendbythecorrespondingdetectorsinthering.Asthesourcerotates,thisprocedureisrepeated.Tomographyconsistsofalgorithmsthatusethesenseddatatoconstructanimagethatrepresentsa"slice"throughtheobject.Motionoftheobjectinadirectionperpendiculartotheringofdetectorsproducesasetofsuchslices,whichconstituteathree-dimensional(3D)renditionoftheinsideoftheobject.TomographywasinventedindependentlybySirGodfreyN.HounsfieldandProfessorAllanM.Cormack,whosharedthe1979NobelPrizeinMedicinefortheirinvention.ItisinterestingtonotethatX-rayswerediscoveredin1895byWilhelmConradRoentgen,forwhichhereceivedthe1901NobelPrizeforPhysics.Thesetwoinventions,nearly100yearsapart,ledtosomeofthemostactiveapplicationareasofimageprocessingtoday.Fromthe1960suntilthepresent,thefieldofimageprocessinghasgrownvigorously.Inadditiontoapplicationsinmedicineandthespaceprogram,digitalimageprocessingtechniquesnowareusedinabroadrangeofapplications.ComputerproceduresareusedtoenhancethecontrastorcodetheintensitylevelsintocolorforeasierinterpretationofX-raysandotherimagesusedinindustry,medicine,andthebiologicalsciences.Geographersusethesameorsimilartechniquestostudypollutionpatternsfromaerialandsatelliteimagery.Imageenhancementandrestorationproceduresareusedtoprocessdegradedimagesofunrecoverableobjectsorexperimentalresultstooexpensivetoduplicate.Inarcheology,imageprocessingmethodshavesuccessfullyrestoredblurredpicturesthatweretheonlyavailablerecordsofrareartifactslostordamagedafterbeingphotographed.Inphysicsandrelatedfields,computertechniquesroutinelyenhanceimagesofexperimentsinareassuchashigh-energyplasmasandelectronmicroscopy.Similarlysuccessfulapplicationsofimageprocessingconceptscanbefoundinastronomy,biology,nuclearmedicine,lawenforcement,defense,andindustrialapplications.Theseexamplesillustrateprocessingresultsintendedforhumaninterpretation.Thesecondmajorareaofapplicationofdigitalimageprocessingtechniquesmentionedatthebeginningofthischapterisinsolvingproblemsdealingwithmachineperception.Inthiscase,interestfocusesonproceduresforextractingfromanimageinformationinaformsuitableforcomputerprocessing.Often,thisinformationbearslittleresemblancetovisualfeaturesthathumansuseininterpretingthecontentofanimage.Examplesofthetypeofinformationusedinmachineperceptionarestatisticalmoments,Fouriertransformcoefficients,andmultidimensionaldistancemeasures.Typicalproblemsinmachineperceptionthatroutinelyutilizeimageprocessingtechniquesareautomaticcharacterrecognition,industrialmachinevisionforproductassemblyandinspection,militaryrecognizance,automaticprocessingoffingerprints,screeningofX-raysandbloodsamples,andmachineprocessingofaerialandsatelliteimageryforweatherpredictionandenvironmentalassessment.ThecontinuingdeclineintheratioofcomputerpricetoperformanceandtheexpansionofnetworkingandcommunicationbandwidthviatheWorldWideWebandtheInternethavecreatedunprecedentedopportunitiesforcontinuedgrowthofdigitalimageprocessing.Someoftheseapplicationareasareillustratedinthefollowingsection.Newwordsandphrasessubmarine[ˈsʌbməri:n;sʌbməˈri:n]n.adj.潜水艇；海底生物海底的；水下的specialized[ˈspɛʃəˈlaizd]adj.专业的；专门的distribution[ˌdistriˈbju:ʃ(ə)n]n.分布；分配reproduction[ˌri:prəˈdʌkʃ(ə)n]n.复制；复制品；繁殖，生殖tape[teip]n.vt.胶带；磁带；带子；卷尺录音；用带子捆扎；用胶布把封住centralprocessingunit(CPU)中央处理机integratedcircuit(IC)集成电路miniaturization[ˌminiətʃəraiˈzeiʃən]n.小型化，微型化verylargescaleintegration(VLSI)超大规模集成电路combination[ˌkɔmbiˈneiʃ(ə)n]n.结合；组合；联合【化学】化合potential[pə(u)ˈtenʃ(ə)l]n.adj.潜能；可能性【电】电势潜在的；可能的；势的distortion[diˈstɔ:ʃ(ə)n]n.变形【物】失真；扭曲；曲解lunar[ˈlu:nə]adj.月亮的，月球的；阴历的；银的；微亮的geometric[ˌdʒiəˈmetrik]adj.几何学的【数】几何学图形的spacecraft[ˈspeiskrɑ:ft]n.【航】宇宙飞船，航天器parallel[ˈpærəlel]n.vt.adj.平行线；对比使……与……平行平行的；类似的，相同的X-ray[ˈeksˈrei]vt.vi.n.adj.用X光线检查使用X光射线；射线照片X光的；与X射线有关的NobelPrize诺贝尔奖biological[baiə(u)ˈlɔdʒik(ə)l]adj.生物的；生物学的restoration[restəˈreiʃ(ə)n]n.恢复；复位；王政复辟；归还archeology[ˌɑ:kiˈɔlədʒi]n.考古学(等于archaeology)blurred[ˈblə:rid,blə:d]adj.模糊不清的；被弄污的microscopy[ˈmaiˈkrɔskəpi]n.显微镜检查；显微镜使用；显微镜学astronomy[əsˈtrɔnəmi]n.天文学nuclear[ˈnju:kliə]adj.原子能的【细胞】细胞核的；中心的；原子核的procedure[prəˈsi:dʒə]n.程序，手续；步骤Fouriertransform傅里叶变换multidimensional[ˌmʌltidiˈmenʃənl]adj.【数】多维的；多面的automatic[ɔ:təˈmætik]adj.自动的；无意识的；必然的aerial[ˈɛəriəl]adj.空中的，航空的；空气的；空想的imagery[ˈimidʒ(ə)ri]n.像；意象；比喻；形象化opportunity[ˌɔpəˈtjuniti]n.机会；机遇Notes1.本节主要介绍数字图像处理的起源及发展。2.

betiedto原意为“束缚于……”，“捆绑于……”，此处翻译为“与……有关”。此句可翻译为：数字图像处理的发展历史和数字计算机的发展密切相关。3.句中的that引导定语从句修饰developments；而在宾语从句中，whatwecallacomputertoday作介词of的宾语，直接译为“计算机”。此句可翻译为：近两个世纪以来算盘的一些发展为计算机的出现奠定了基础。4.

datebackto追溯到。此句可译为：然而，现代计算机的基础还要回溯到20世纪40年代由约翰·冯·诺依曼提出的两个重要概念：(1)保存程序和数据的存储器；(2)条件分支。5.

Inparallelwith与……同时，与……平行。此句可译为：进行空间应用的同时，数字图像处理技术自20世纪60年代末和20世纪70年代初开始还被用于医学图像、地球遥感监测和天文学等领域。6.句中inwhich引导定语从句修饰名词process，因为该定语从句较长，翻译时采用分译法，即将该定语从句译为一个句子，而不将其译为process的前置定语。此句可译为：计算机轴向断层术是一种处理方法，在这种处理中，一个检测器环围绕着一个物体(或病人)，并且一个X射线源(与检测器环同心)绕着物体旋转。9.3FundamentalStepsinDigitalImageProcessing1ItishelpfultodividethematerialcoveredinthefollowingsectionsintothetwobroadcategoriesdefinedinSection9.1:methodswhoseinputandoutputareimages,andmethodswhoseinputsmaybeimages,butwhoseoutputsareattributesextractedfromthoseimages2.ThisorganizationissummarizedinFigure9.2.Thediagramdoesnotimplythateveryprocessisappliedtoanimage.Rather,theintentionistoconveyanideaofallthemethodologiesthatcanbeappliedtoimagesfordifferentpurposesandpossiblywithdifferentobjectives.Figure9.2Fundamentalstepsindigitalimageprocessing

ImageacquisitionisthefirstprocessshowninFiguere9.2.ThediscussioninSection9.2gavesomehintsregardingtheoriginofdigitalimages.Thistopicisconsideredinmuchmoredetailinthefollowing,wherewealsointroduceanumberofbasicdigitalimageconceptsthatareusedthroughoutthechapter.Notethatacquisitioncouldbeassimpleasbeinggivenanimagethatisalreadyindigitalform.Generally,theimageacquisitionstageinvolvespreprocessing,suchasscaling.Imageenhancementisamongthesimplestandmostappealingareasofdigitalimageprocessing.Basically,theideabehindenhancementtechniquesistobringoutdetailthatisobscured,orsimplytohighlightcertainfeaturesofinterestinanimage3.Afamiliarexampleofenhancementiswhenweincreasethecontrastofanimagebecause"itlooksbetter".Itisimportanttokeepinmindthatenhancementisaverysubjectiveareaofimageprocessing.Imagerestorationisanareathatalsodealswithimprovingtheappearanceofanimage.However,unlikeenhancement,whichissubjective,imagerestorationisobjective,inthesensethatrestorationtechniquestendtobebasedonmathematicalorprobabilisticmodelsofimagedegradation4.Enhancement,ontheotherhand,isbasedonhumansubjectivepreferencesregardingwhatconstitutesa"good"enhancementresult.ColorimageprocessingisanareathathasbeengaininginimportancebecauseofthesignificantincreaseintheuseofdigitalimagesovertheInternet.Compression,asthenameimplies,dealswithtechniquesforreducingthestoragerequiredtosaveanimage,orthebandwidthrequiredtotransmitit5.Althoughstoragetechnologyhasimprovedsignificantlyoverthepastdecade,thesamecannotbesaidfortransmissioncapacity.ThisistrueparticularlyinusesoftheInternet,whicharecharacterizedbysignificantpictorialcontent.Imagecompressionisfamiliar(perhapsinadvertently)tomostusersofcomputersintheformofimagefileextensions,suchasthe.jpgfileextensionusedintheJPEG(JointPhotographicExpertsGroup)imagecompressionstandard.Morphologicalprocessingdealswithtoolsforextractingimagecomponentsthatareusefulintherepresentationanddescriptionofshape.Thematerialinthischapterbeginsatransitionfromprocessesthatoutputimagestoprocessesthatoutputimageattributes,asindicatedinSection9.1.Segmentationprocedurespartitionanimageintoitsconstituentpartsorobjects.Ingeneral6,autonomoussegmentationisoneofthemostdifficulttasksindigitalimageprocessing.Aruggedsegmentationprocedurebringstheprocessalongwaytowardsuccessfulsolutionofimagingproblemsthatrequireobjectstobeidentifiedindividually.Ontheotherhand,weakorerraticsegmentationalgorithmsalmostalwaysguaranteeeventualfailure.Ingeneral,themoreaccuratethesegmentation,themorelikelyrecognitionistosucceed.Representationanddescriptionalmostalwaysfollowtheoutputofasegmentationstage,whichusuallyisrawpixeldata,constitutingeithertheboundaryofaregion(i.e.,thesetofpixelsseparatingoneimageregionfromanother)orallthepointsintheregionitself7.Ineithercase,convertingthedatatoaformsuitableforcomputerprocessingisnecessary.Thefirstdecisionthatmustbemadeiswhetherthedatashouldberepresentedasaboundaryorasacompleteregion.Boundaryrepresentationisappropriatewhenthefocusisonexternalshapecharacteristics,suchascornersandinflections.Regionalrepresentationisappropriatewhenthefocusisoninternalproperties,suchastextureorskeletalshape.Insomeapplications,theserepresentationscomplementeachother.Choosingarepresentationisonlypartofthesolutionfortransformingrawdataintoaformsuitableforsubsequentcomputerprocessing.Amethodmustalsobespecifiedfordescribingthedatasothatfeaturesofinterestarehighlighted.Description,alsocalledfeatureselection,dealswithextractingattributesthatresultinsomequantitativeinformationofinterestorarebasicfordifferentiatingoneclassofobjectsfromanother.Recognitionistheprocessthatassignsalabel(e.g.,"vehicle")toanobjectbasedonitsdescriptors.AsdetailedinSection9.1,weconcludeourcoverageofdigitalimageprocessingwiththedevelopmentofmethodsforrecognitionofindividualobjects.SofarwehavesaidnothingabouttheneedforpriorknowledgeorabouttheinteractionbetweentheknowledgebaseandtheprocessingmodulesinFigure9.2.Knowledgeaboutaproblemdomainiscodedintoanimageprocessingsystemintheformofaknowledgedatabase.Thisknowledgemaybeassimpleasdetailingregionsofanimagewheretheinformationofinterestisknowntobelocated,thuslimitingthesearchthathastobeconductedinseekingthatinformation.Theknowledgebasealsocanbequitecomplex,suchasaninterrelatedlistofallmajorpossibledefectsinamaterialsinspectionproblemoranimagedatabasecontaininghigh-resolutionsatelliteimagesofaregioninconnectionwithchange-detectionapplications.Inadditiontoguidingtheoperationofeachprocessingmodule,theknowledgebasealsocontrolstheinteractionbetweenmodules.ThisdistinctionismadeinFigure9.2bytheuseofdouble-headedarrowsbetweentheprocessingmodulesandtheknowledgebase,asopposedtosingle-headedarrowslinkingtheprocessingmodules.Althoughwedonotdiscussimagedisplayexplicitlyatthispoint,itisimportanttokeepinmindthatviewingtheresultsofimageprocessingcantakeplaceattheoutputofanystageinFigure9.2.WealsonotethatnotallimageprocessingapplicationsrequirethecomplexityofinteractionsimpliedbyFigure9.2.Infact,notevenallthosemodulesareneededinsomecases.Forexample,imageenhancementforhumanvisualinterpretationseldomrequiresuseofanyoftheotherstagesinFigure9.2.Ingeneral,however,asthecomplexityofanimageprocessingtaskincreases,sodoesthenumberofprocessesrequiredtosolvetheproblem.Newwordsandphrasesdiagram[ˈdaiəɡræm]n.vt.图表；图解用图解法表示methodology[ˌmeθəˈdɔlədʒi]n.方法学，方法论imageacquisition图像获取preprocessingn.预处理；预加工scaling[ˈskeiliŋ]n.缩放比例；刮治术【口腔】刮牙术；鳞片排列imageenhancement图像增强highlight[ˈhailait]vt.n.突出；强调；使显著；加亮最精彩的部分；最重要的事情contrast[ˈkɔntra:st]n.vi.vt.对比，对比度；差别对比；形成对照使对比；与……对照imagerestoration图像复原subjective[sʌbˈdʒektiv]adj.主观的；个人的；自觉的probabilistic[ˌprɔbəbiˈlistik]adj.概率性的；或然说的，盖然论的Degradation[ˌdegrəˈdeiʃ(ə)n]n.退化；降格，降级；堕落morphological[ˌmɔ:fəˈlɔdʒikəl]adj.形态学的representation[ˌreprizenˈteiʃ(ə)n]n.代表；表现；表示法；陈述attribute[əˈtribju:t]n.vt.属性；特质归属；把……归于segmentationprocedure分割过程partition[pɑ:ˈtiʃ(ə)n]vt.n.【数】分割；分隔；区分划分，分开【数】分割；隔墙；隔离物erratic[iˈrætik]adj.不稳定的；古怪的boundary[ˈbaund(ə)ri]n.边界；范围；分界线inflection[inˈflekʃ(ə)n]n.弯曲，变形；音调变化texture[ˈtekstʃə]n.质地；纹理；结构；本质，实质skeletal[ˈskɛlətl]adj.骨骼的，像骨骼的；骸骨的complement[ˈkɔmplim(ə)nt]n.vt.补足物；补语；余角补足，补助quantitative[ˈkwɔntiˌtətiv]adj.定量的；量的，数量的interaction[intərˈækʃ(ə)n]n.n.相互作用【数】交互作用互动priorknowledge先验知识module[ˈmɔdju:l]n.【计】模块；组件；模数explicitly[ikˈsplisitli]adv.明确地；明白地complexity[kəmˈpleksəti]n.复杂性；复杂错综的事物Notes1.本节主要介绍数字图像处理所包含的内容以及基本步骤。2.句中it是形式主语，真正的主语是helpful后的动词不定式todivide。此句可翻译为：把后续各部分涉及的材料划分为在9.1节定义的两个主要类别是有帮助的。一类的输入和输出都是图像；另一类的输入可能是图像，但是输出是从图像中提取的特征属性。3.句中连词or连接动词不定式tobringout和tohighlight作表语。此句可翻译