Unit3DataStorageandCompression23-1BitPattern3-2IntegerinComputer3-3Floating-PointinComputer3-4DataStorage3-5CloudStorageandBigDataStorage3-6ReferencesandRecommendedReading3-7Summary3-8PracticeSetOUTLINE3Afterreadingthischapter,youaresupposedto

beableto:OBJECTIVESUnderstandthedifferentrepresentationsofanintegerinside

acomputer:unsigned,sign-and-magnitude,one’scomplement,

andtwo’scomplement.UnderstandtheExcesssystemthatisusedtostoretheexponentialpartofafloating-pointnumber.Understandhowfloatingnumbersarestoredinsideacomputer

usingtheexponentandthemantissa.Understandtheuniformrepresentationinsidethecomputerfordifferentdatatypes.4Understandhowdifferenttypesofdataarerepresentedandcompressedinsidethecomputer.OBJECTIVESUnderstandthenewnetworkstoragetechnology–cloudstorage.Understandthemainstoragetechnologiesforbigdata.53-1BitPattern

Notsolongago,computerswereusedtodoscientificcalculationsanddisplayhuman-readabletext.Butnowtheyareactuallymultimediadevices,dealingwithmanydifferenttypesofdata,includingnumbers,text,audio,images,andvideo.

Forefficiency,allaforementioneddatatypeshaveauniformrepresentationinsidethecomputer,whichisbasedonabinarysystem,calledabitpattern【位模式/位組】orbinaryrepresentation【二進(jìn)制表示】.6BitPatternInfact,therehavebeencomputersbasedondecimalnumbersystemorothernumbersystems.Butthesecomputersaretoocomplicated,expensive,andlessstableforeverydayuse.Intuitively,ifasinglelogicunitrepresentslesspossiblevalues,thecomputerwillbemore

stable.Foranelectronicsignal,thebinarysystemprovidesonlytwopossiblestates:loworhigh,correspondingto0or1respectively.Anelectronicswitchcanbeeasilyrepresentedbytwostablestatesofelectronicsignal.Thesesatescanbedefinedasonandoffrepresentedby1and0respectively.

Torepresentmorethantwothings(onandoff),onebitisnotenough,soweneedasequenceofbits.Ingeneral,mbitsarecapableofrepresenting2m

thingsbecausembitscanmake2mcombinationsof0and1.Computermemoryhasnoideawhattypeofdataastoredbitpatternrepresents.7BitPattern

Computermemory【內(nèi)存】hasnoideawhattypeofdataastoredbitpatternrepresents.Normally,thememorystoresvariouspiecesofdatawithdifferentpatternsaccordingtotheirdatatypes【一般情況下，內(nèi)存使用不同的模式，根據(jù)數(shù)據(jù)類型，存儲(chǔ)多份數(shù)據(jù)】,asshowninTable3.1.Forexample,inatexteditor,thecharacter‘#’canbestoredas00100011,andthenumber‘38’canberepresentedwiththe8-bitpattern00100110inamathematicalroutine.8Table3.1DatatypesProgramMemoryAnumber‘38’Mathroutine00100110Acharacter‘#’Texteditor00100011PartofanimageImagerecorder10100110PartofasongMusicrecorder10110110PartofafilmVideorecorder1010011193-2IntegerinComputerIntegersrepresentnumberswhichhavenofractionalpart.AsdiscussedinChapter2,eachbitpatterncanbetreatedasabinarynumber,andeachbinarynumbercanbeconvertedtothecorrespondingdecimalinteger.It’snaturaltouseabitpatterntorepresentintegers.Afixed-pointrepresentation【定點(diǎn)表示法】

isthemethodusedforstoringintegersinbinaryformat.Thedecimalpointisassumedattherightoftheleastsignificant(rightmost)bit【最低有效位；最右位】withoutanygap.10IntegerinComputerTomakemoreefficientuseofcomputermemory,unsignedinteger【無(wú)符號(hào)整數(shù)】andsignedinteger【有符號(hào)整數(shù)】havebeendevelopedastwocategoriesofintegerrepresentation.Threedistinctwayshavebeendevelopedtorepresentsignedintegers：

Sign-and-Magnitude【原碼】;One’sComplement【二進(jìn)制反碼/一的補(bǔ)碼】;andTwo’sComplement【二進(jìn)制補(bǔ)碼】11UnsignedrepresentationSignedrepresentationIntegerinComputer12Unsignedrepresentationisusedtorepresentpositiveintegersandzero.Storingunsignedintegersisastraightforwardprocess,whichisalmostthesameastheproceduresshowninChapter2.

!!!!Afterthenumberischangedtobinaryofn

bits,thecomputeradds(m-n)

0stotheleftwhennissmallerthanthelengthofmemorylocationm.Forinstance,an8-bitmemorylocationusesunsignedrepresentation00001111tostore15,herefour(8-4)0sareaddedtheleftmostof1111.Notethattheleftleading0sareessential.UnsignedRepresentation13SignedRepresentationUnlikepositiveintegers,negativeintegersarerepresentedwithaminussigninfrontofthem.Torepresentnegativeintegersinbinarynumbersystems,signedrepresentationhasbeendeveloped.Inspiredbythebinarysystem,asigncanbeindicatedbyonebitwhichhasexactlytwovalues.Alloftheabovethreesignedrepresentationformatsuse0forpositivesignand1fornegativesignastheleftmostbit.Forapositiveinteger,itsconventionisthesameasanunsignedrepresentation,andthesignbitequals0.Sotherangeofpositiveintegersrepresentedbyanunsignedrepresentationisdifferentfrompositiveintegersrepresentedbyasignedrepresentation.Thelatter’srangeisjusthalfoftheformerone.Thethreesignedrepresentationstoencodenegativeintegersareslightlydifferentwhiletheytoencodepositiveintegersareidentical14SignedRepresentationSign-and-magnitudeOne’scomplementrepresentationTwo’scomplementrepresentation15Sign-and-magnitudeSign-and-magnituderepresentationissimple.Wecantransformitsmagnitudeparttodecimalvaluedirectly.Thefollowingfigureshowstherangeofsign-and-magnituderepresentationwith5-bitallocation.Sinceonly4bitscanbeusedtorepresentthemagnitudepart,thisrangeisdividedintotwohalves:00000to01111and10000to11111.Theformerrepresentspositiveintegers,andthelatterrepresentsnegativeintegers.Youmaynoticetherearetwozeros:+0and-0.Weoftenignorethenegativezeroentirely,butwithinthecomputer,itmaycauseunnecessarycomplexity.It’seasytoconcludethatanm-bitlocationcanrepresentnumbersfrom-(2m-1-1)to+(2m-1-1)16One’sComplementRepresentationOne’scomplementrepresentationofanegativebinaryintegeristhecomplementofitspositivecounterpart.Foranumber,itscomplementisobtainedbychangingallbitsof0sto1sandallbitsof1sto0s.

Therangeofsignedintegersusingone’scomplementinacomputeristhesameastherangeofsign-and-magnitudeintegers.Therearealsotworepresentationsof0inthisformat,denotedby-0(negativezero)and+0(positive0).17Example1Store–268ina16-bitmemorylocationusingone’scomplementrepresentation.SolutionFirst,thenumbercanbechangedtobinary100001100.Addseven0stomakeatotalofN(16)bits,0000000100001100.Thesignisnegative,soeachbitisinvertedtogetthenumberinone’scomplement.Theresultis:

11111110111110011NB:theresultistotallydifferentfromthatforthesign-and-magnituderepresentation[1000000100001100]!18-268inone’scomplementformat.One’sComplementRepresentation19Forone’scomplementrepresentation,an8-bitmemorylocationcannotstoreapositiveintegerlargerthan255oranegativeintegerlessthan-255,otherwiseoverflowwouldoccur.Becauseofnegativezeroandend-aroundborrow【循環(huán)借位】(whichmaybeproducedbyone’scomplementsubtraction),one’scomplementrepresentationisnotcommonlyusednow.One’sComplementRepresentation20Two’sComplementRepresentationInfact,two’scomplementisthemostcommonmethodusedbycomputertorepresentandstoreasignedintegerwithm-bit.Anunsignedintegerof(0to2m?1)usingtwo’scomplementrepresentationalsohastwoavailablesub-ranges.

Unliketheprevioustwomethodsforsignedintegers,two’scomplementhasonlyonerepresentationofzero,and1000isaspecialcasewhoseleftmostbit1notonlyrepresentanegativesign,butalsocontributetothedecimalvalue.Sotherangeofnumbersthatm-bitlocationcanstoreusingtwo’scomplementis-2m-1to+(2m-1-1).21Two’sComplementRepresentationForpositiveinteger,theconversionfromdecimaltobinaryisenough,justlikebefore.Notethattheleftpadding0sarestillneededifbitsarenotenough.Iftheintegerisnegative,theoperationincludestwosteps.First,weleavetherightmost0suptothefirst1(includingthe1)unchangedandthencomplementtherest.Thetwo’scomplementofanintegercanalsobecompletedbyadding1totheresultaftertakingthecomplement.22Example2Store–5ina4-bitmemorylocationusingtwo’scomplementrepresentation.SolutionFirst,ignorethesignofnumber5,changeittobinary

101.

Addone0tomakeatotalofN(4)bits,

0101.Thesignisnegative,soleavetherightmost0suptothefirst1(includingthe1)unchangedandcomplementtherest(orcomplement0101=1010,thenadd1),Theresultis:

101123-5intwo’scomplementformat.Two’sComplementRepresentation24Two’sComplementRepresentationThefollowingtableshowsthedifferencesamongalltheseintegerrepresentations.Inthistable,weassumethatmis4,sothememorylocationcanstoreonly4bits.Notethattheinterpretationisdifferentfornegativeinteger,whileitisthesameforpositiveinteger.25SummaryofIntegerRepresentationsContentsofMemoryUnsignedSign-and-magnitudeOne’scomplementTwo’scomplement00000+0+0+000011+1+1+100102+2+2+200113+3+3+301004+4+4+401015+5+5+501106+6+6+601117+7+7+710008-0-7-810019-1-6-7101010-2-5-6101111-3-4-5110012-4-3-4110113-5-2-3111014-6-1-2111115-7-0-1263-3Floating-PointinComputerAnumberwhichconsistsofanintegralpartandafractionalpartiscalledareal【實(shí)數(shù)】.Althoughwecanusefixed-pointrepresentationtorepresentarealnumber,however,theresultishardtobeaccurateenoughbecausethedigitsinintegralpartandfractionalpartarebothfixed.Thesolutionistousefloating-pointrepresentation.Floating-pointrepresentation【浮點(diǎn)表示法】adoptsscientificnotation,withonlyonenon-zerodigittotheleftofthedecimalpoint“.”.Inthebinarysystem,thisdigitcanonlybe1.27Asshowninthefollowingfigure,theprocessofconvertingareal(101100010.01inthefollowingexample)toscientificnotationiscallednormalization.Afternormalization,onlysign,exponentandmantissaneedtobestored.Notethattheexponentbase2,leftmostbit1,anddecimalpointinthemantissaareimplicit.Floating-PointinComputer28Floating-PointinComputerConvertingthefractionpartofarealnumbertobinaryissimilartochanginganaturalnumberfrombase10tobase2,butinsteadofdividing,itmultiplythefractionpartbythebase.Thewholenumberpartoftheresultisthefirstbinarydigittotherightofthepoint.Next,thefractionalpartofthepreviousresultismultipliedbybase2.Thestepsarerepeateduntilthefractionalpartiszerooruntilaninfiniterepeatingpatternisrecognized,andthen,wehavetostopaftertherequiredprecisionisreached.29Toconvert0.25tobinary,multiplythefractionby2;theresultis0.50.Theintegerpartoftheresult(0)isextractedandbecomestheleftmostbinarydigit.Nowmultiplyby2thefractionpart(0.50)oftheresulttoget1.00.Again,theintegerpartoftheresult(1)isextractedandbecomesthenextbinarydigit.0.250.501.00.0

0.01Convertingthefractionpart30ExampleTransformthefraction0.815tobinarySolutionWritethefractionattheleftcorner.Multiplythenumbercontinuouslyby2andextracttheintegerpartasthebinarydigit.Stopwhentheprecisionisreached.0.8151.6301.2600.520

0.1101.0400.0800.160100Pleasenoticethatwehavetostopasitseemsthatanirregularinfiniterepetitionisoccurring31Floating-PointinComputerNowlet’sgothroughtheentireconversionprocessbyconverting30.75fromdecimaltobinary.Firstweconvert30.Therefore,30is11110inbinary(placedfromrighttoleft!).

32Floating-PointinComputerThenwechangethefractionalparttobinary:Therefore,30.75indecimalis11110.11inbinary.

.75*2=1.50.50*2=1.0033Floating-PointinComputerIEEE(InstituteofElectricalandElectronicsEngineers)FPS

(Floating-PointStandard)hasdefinedtwostandardsforstoringfloating-pointnumbersinmemory:singleprecision(32bits),anddoubleprecision(64bits).Floating-pointnumbers’standardsSignExponentMantissasingleprecision(32bits)1bit8bits23bitsdoubleprecision(64bits)1bit11bits52bits34Floating-PointinComputerTheIEEEstandarddefines1bitforsign,either0or1.Itdefinesfloatingpointnumbersintheirnormalizedform.Tonormalizethemantissa,IEEEstandardnormalizethefixed-partofarealanduseunsignedintegerrepresentationtostoreit.Thenumberofdigitsusedtoshiftthedecimalpointtoleftorrightisindicatedbytheexponent,andthisnewrepresentationiscalledexcessrepresentation【移碼】.35ExcessrepresentationExcessrepresentationisalsoknownasbiasedrepresentation.Itaddsadesignatedbiasedvalue(ormagicnumber)totheoriginalvaluetostoreallexponentsasanunsignedinteger.Iftheexponentoccupiesmbitsincomputermemory,thedesignatedbiasedvalueis2m?1_1(referredtoasL).Theshiftinginexcesssystemwith4-bitallocationisshowninthefollowingfigure.ThisnewsystemisgenerallycalledExcess-L,likeExcess-7.36Floating-PointinComputerForinstance,weuseIEEE754singleprecisionformattorepresent-281.875.Forsingleprecision,thenumberisdividedintosignbit,exponent,andfraction(alsocalledsignificand【有效數(shù);尾數(shù)】ormantissa).Theexponentisencodedasan8-bitpattern,sothebiasis127(orExcess-127).Weproceedasfollows:(1)

Thesignisnegative,sovalueofsignbitis1,thatisS=1.(2)Transform281.875todecimal:(100011001.111)2.(3)Normalization:(100011001.111)2=(1.00011001111)2×28.(4)EistheexponentfieldandMisthemantissa.E=8+127=135=(10000111)2andM=(00011001111)2.Weneedtoadd12zerostotherightofMtomakeit23

bits.37Floating-PointinComputerThefinalrepresentationisshown:Carefulreadermaynoticethatnumberswithtoosmallortoolargeabsolutevaluescannotbestoredwiththisrepresentation.Tohandlethisspecialcase,itisagreedthatthevalueequals0whenthesign,exponentandmantissaareall0bits.Excess-127givesanexponentrangeof2-126to2+127,andexponent0(2-127)and255(2+128)arereservedforspecialcases.38Aspreviouslydescribed,numbers,text,images,audioandvideoarefivedifferenttypesofdata.InSection3.2andSection3.3,themethodsofstoringnumbersinsideacomputermemoryhavebeenintroduced.Inthissection,wewilldiscusstherepresentationforotherformsofdata,normallycalleddataencoding.Inordertoreducetheamountofstoragespace,theredundancyofdataneedstobeeliminated,andthetechniqueusediscalleddatacompression.3-4DataStorage39StoringtextStoringaudioStoringimagesStoringvideo

DataStorage40StoringTextAtextdocumentiscomposedofvariouscharacters.IntheEnglishlanguage,thereare26letters.Butconsideringtheuppercaseandlowercase,therearetotal52uniquecharactersthatneedtobetreatedseparately.Besides,10numericcharacters,variouspunctuationcharacters,tabcharacter,spacecharacter,andnewlinecharacter,etc.alsohavetoberepresented.Acharacterencodingisthemappingbetweenthesetofcharactersandthecodesusedtorepresentthem.TwoofthemostwidelyusedcharactersetsASCII(AmericanStandardCodeforInformationInterchange)andUnicodeareintroducedbelow.ASCII【美國(guó)信息交換標(biāo)準(zhǔn)碼】andUnicode【統(tǒng)一碼】Run-lengthencoding【行程長(zhǎng)度編碼】Huffmanencoding【哈夫曼編碼】41ASCIIandUnicode(1)ASCII

ASCIIuses7bitstorepresent128uniquecharacters.ASCIIreservesthefirst32codesforcontrolcharacters,whicharedifferentfromthefollowingprintablecharacters(x20tox7E).It’seasytogetafullASCIItablefromtheInternet.ButevenextendedASCII(8bits)isnotenoughforothercommonlanguages,suchasChinese,whichneedssymbolsfarbeyondthe256charactersofthestandardASCIIcharacterset.(2)UnicodeTorepresenteverycharacterintheworldforinternationaluse,theUnicodecharactersetwasdeveloped.Unicodeuses16-bit(torepresentover65thousandcharacters(65,536))andincludesallthecharactersoftheextendedASCIIasitsfirst256characters.Today,Unicodeiswidelyusedbymanycomputersystemsandprogramminglanguages.42ASCII43ASCII44Run-lengthencodingRun-lengthencodingisintendedtocompressthetextwithsomecharactersrepeatingoverandoveragaininalongsequence.Thegeneralideaistoreplaceasequenceofrepeatedcharacterswiththerepeatedcharacterfollowedbytherepetitioncount.Toindicatethestartoftherepeatedcharacters,aflagcharacterisneeded.45Example3Usingrun-lengthencodingwithflagcharacter‘#’torepresentGGGGGAAAAAACCCC(15)SolutionTheresultis#G5#A6#C4(9).Thenumberfollowingthecharacterspecifiesthenumberoftimesthecharacterisappearing.Thenewcompressionsequencecontains9charactersinsteadof15intheoriginaltext46HuffmanencodingHuffmanencodingisanothertextcompressiontechnique,basedonthefactthatsomecharactersoccurmorefrequentlythanothers.Theideaistouseshorterbitstringstoreplacethemorefrequentcharactersandleavelongerbitstringstothoseappearinglessfrequently,47Huffmanencodingthebitstringrepresentingsomeparticularsymbolshouldneverbeaprefixofthebitstringrepresentinganyothersymbol.Forsimplicity,supposeadocumentonlycontainsfivekindsofcharacters:A,B,C,D,E,theencodingtableislistedinTable3.4,andthefulltextisasfollows:48HuffmanencodingAEBEAECADEAEBEABECCCADDDDDDBAECCAEBCEBAECCADDDDDDEAEBDDCEEEEAECADEEEEEDDDDBDCEAEBEEDDDCDDBCEAECDDADE.AsshowninTable3.4,characterCoccurs15times,characterDappears29times,etc.49HuffmanencodingTogeteachcharacter’sHuffmancode,itisnecessarytobuildupatreefromthebottomup.Itsstepsisasfollows:first,createaleafnodeforeachcharacter,andputallthenodesinarow;second,combinetwosmallestfrequencynodesintoanewinternalnodewiththesetwonodesaschildren,andthefrequencyofthenewnodeisequaltothesumofthesetwonodes’frequencies;third,repeatsecond50Huffmanencodingstepuntilallthenodesareinasingletree.Then,usethetreetoassigncodestoeachcharacter,assign1totherightbranchand0totheleftbranchanddothesameateachnode.Last,startfromtherootnodeandfollowthebranchestoeachcharacter,theHuffmancodesofallcharactersarefound,asshownincolumn3ofTable3.4.51Example4Thus,withHuffmancode,thefirst10characters:AEBEAECADEcanbeencodedas0011010110011011001011.Theoriginallengthofitis10*8bits=80bitsusingASCIIcode.AfterHuffmanencoding,thelengthdecreasesto22bits,withacompressionratioof0.275(22/80).SolutionCharacterFrequencyHuffmanCodeA1600[2bits]B10010[3bits]C15011[3bits]D2910[2bits]E3011[2bits]52Storingaudio

Wecandividethedatatypesintotwocategories:analogdataanddigitaldata.Analogdataisrealworldstufflikesounds,paintings,andtemperature,whichisnotcountable.Digitaldata,ontheotherhand,isdiscrete,andcountable,likecharactersintext.53(1)DigitizationAudioisakindofanalogdata,whilecomputerscanonlyhandledigitaldata,soaudiodatahastobesampled【采樣】andthenconvertedtonumericvalues.Thisprocessissometimescalled“digitize”.torepresentthesoundwave,afinitenumberofsamplepointsonthatcurvearenecessary.Thenumberofsamplepointspersecondiscalledsamplingrate.Foreachsample,quantizationistheprocessthatroundsthemeasuredvaluetotheclosestintegervalue.Beforebeingstoredinthecomputer,eachvalueshouldbeconvertedtoabitpattern.Thebitlengthforeachsamplevalueissometimescalledbitdepthorwordlength.Themultiplicationofsamplingratebythebitdepthisnormallycalledbitrate.Digitization54(2)EncodingformatThereareseveralpopularformatsforaudioencoding,likeWAV(Wave),MP3(MPEG-2AudioLayer3)MP4

(MPEG-4Part14),WMA(WindowsMediaAudio)andsoon.Currently,MP3isadominantstandard.ThisstandardisapartofMPEG-1(MotionPictureExpertsGroup)【動(dòng)態(tài)圖像專家組】whichisusedforcompressingvideodata.Ifhumancouldnotheartheinformation,thenitwillbeignoredbythecompressionmethod.Therearealsoseveralstandards,suchasGSM(GlobalSystemforMobileCommunications)G.729andG.723.3.Besides,encodingstandardsforAndroidoriPhonelikeAMR(AdaptiveMulti-Rate)oriLBC(InternetLowBitrateCodec)arewidelyusedtotransfervoicethatmanypopularapplicationsadopt.Encodingformat55

Techniquesusedbycomputerstostoreimagescontainvectorgraphics【矢量圖形】andrastergraphics【光柵圖形】.Rastergraphicsisusedtostorephotographs,whichisananalogrepresentationofanimage.Aphotographalsoneedstobedigitized.Printedphotographstakenwithadigitalcameramayalsobedigitized.Foraudio,theintensityofdatavariesintime,butforimages,itvariesinspace.Scanningisusedasasamplingmethodforimages.Itssamplesarenormallycalledpixels

【像素】andthenumberofpixelsusedtorepresentthephotoiscalledresolution【分辨率】.Storingimages56Withhighenoughresolution,thediscretepixelscanbecontinuoustohumaneye.Asseveralencodingtechniquescanbeusedtohandlethepixel’scolor,apixelcanberepresentedbydifferentnumbersofbits,whichiscalledcolordepth【色深】.24bits(whicharethree8-bits,forRed,Green,andBluerespectively)areusedbyTrue-Colorschemetoencodeasinglepixel,andJPEG(JointPhotographicExpertsGroup【聯(lián)合圖像專家組】)usesthisschemetoencodeimageandthencompressesittoconsumelessbits.Indexedcolorscheme【顏色指數(shù)方案】onlyusesaportionofthesecolors,andisusedbyGIF(GraphicInterchangeFormat【圖形交換格式】)toencodeimage.Storingimages57Mainimageformats(PartA)ImageformatBriefdescriptionCharacteristicsApplicationsceneExtensionnameBitMaPBMP(BitMaP)isthestandardimagefileformatusedinWindowsoperatingsystem,whichusesbitmapasstorageformat.Inadditiontooptionalimagedepth,itdoesnotuseanyothercompressiontechniques.Support1-24colordepth.ImagesoftwarerunningonWindows.BMPPersonalComputereXchangePCXwasdevelopedbyZSOFTinthedevelopmentofimageprocessingsoftwarePaintbrush.It’saproprietaryformatforPC-baseddrawingprogram,andthegeneraldesktoppublishing,graphicartsandvideocapturesoftwaresupportthisformat.Run-lengthencoding.PC-baseddrawingprograms.PCX58Mainimageformats(PartB)ImageformatBriefdescriptionCharacteristicsApplicationsceneExtensionnameTagImageFileFormatTIFFwasagenericimagefileformatdevelopedbyAldusandMicrosoftforthedesktoppublishingsystem.Supportsmultipleencodingmethods.Desktoppublishingsystem,GIS,andremotesensing.TIFFTaggedGraphicsTGAwasdevelopedbyTruevisionforitsgraphicscard.Ithasbeenacceptedbytheinternationalgraphicimageindustry.SupportsIrregularlyshapedgraphics.Thefieldofmultimedia.TGA59Mainimageformats(PartC)ImageformatBriefdescriptionCharacteristicsApplicationsceneExtensionnameGraphicsInterchangeFormatGIFwasdevelopedbyCompuServein1987,itscompressionrateisgenerallyabout50%,andalmostallsoftwaresupportit.Cansavemultiplecolorimages.TheInternet,simpleanimation.GIFJointPhotographicExpertGroupJPEGisthenetwork'smostpopularimageformat,developedbytheJointPhotographicExpertsGroup.Itisalossycompressionformat,andcancompressanimageinasmallstoragespace.JPGisshortforJPEG,andjpgisasuffix,jpegcanbeusedasasuffixortorepresentafileformat.Variablecompressionratio.TheInternet.JPEG60Mainimageformats(PartD)ImageformatBriefdescriptionCharacteristicsApplicationsceneExtensionnameExchangeableImagefileFormatEXIFwaspromotedbyFujifordigitalcamerain1994,itiscapableofstoringphotographicdate,theuseofaperture,flashexposuredataandotherinformation.Storesexposuredata,likephotographydate.DigitalCameras.EXIFkodakFlashPiXFPXwasjointlydevelopedbyKodak,Microsoft,HP,andLivePicture,whichhasmulti-resolution.Withmultipleresolution.UsedbythePictureEasySoftwareapplicationincludedwithKodakdigitalcameras.FPX61Mainimageformats(PartE)ImageformatBriefdescriptionCharacteristicsApplicationsceneExtensionnameScalableVectorGraphicsItisbasedonXML(ExtensibleMarkupLanguage),developedbytheWorldWideWebConsortium's.Anditcanbearbitrarilyenlargedwhilekeepingveryclearedge.CanenlargeGraphicarbitrarily.DesigningWebgraphicspagesofhighresolution.SVGkodakPhotoCDPCDisaPhotoCDfileformatdevelopedbyKodak.TheformatusesYCCcolormodetodefinecolorsintheimage.UsesYCCcolormode.SavepicturesonCD-ROM.PCD62Mainimageformats(PartF)ImageformatBriefdescriptionCharacteristicsApplicationsceneExtensionnamePhotoShopDocumentPSDisaproprietaryfileformatf