The Helmholtz–Kohlrausch effect_图文

ReviewPaper:TheHelmholtz–io(SIDSeniorMemeber)Abstract—TheHelmholtz–Kohlrausch(H–K)effect isthe influenceofcolorpurityontheperceivedbrightnessofacolorobject(orsource).Inadditiontoareviewoftheeffect,asurveyofcolorandbrightness-perceptionstudies from1825to thepresent(includingourownstudies ondisablingglare)is ingglareis theblindingexperiencewhichresults from abrightsourceinourfieldofview.1TherehasbeenagreatdealofworkontheH–Keffect,–Keffecthasleadtonewcolor-appearancemodelswhichhelptodescribethebrightness/luminancediscrepancies,beshownhowtheH–Keffectaffectsmanyofthedisplays,mobiledevices,onally,howtheH–Keffectisinvolvedwithourperceptionatds—Helmholtz-Kohlrauscheffect,mesopic,#10.1889/JSID19.10.6581IntroductionThatbrightness(the sensation)doesnot alwaysvarylinearlywithluminance(thestimulus),,light, sound,orelectric shock)isa powerlaw2,3andattimesreferredtoasalogfunction.4Wearedealingwiththebrightnesseffectsthatoccurwhenthe colorofalightsource(orobject) addressthecaseswheretheluminanca5explainstheluminance–brightnessdiscrepancythisway:“brightnessisaconceptofcolorimetryandnotofphotometry.”Thus,oneerontheH–Keffectdescribeshowthebrighreaseinsourcebrightnessasthepurityofthelightsourceincreases(orthespectrallinewidthofthelightsourcedecreases)iscentraltotheH–epthecolorsthesameandchangetheambientlight,wehaveanotherbrightness–luminancediscrepancy,thePurkinjeeffect,whichinvolvesthetransitionbetweenphotopicandscotopicvioingtointroducethereadertotheHelmholtz–Kohlrausch(H–K)effect,whichalsoshowsa typeofbright-ness–peris,inpart,histori-calandthusisachronologicaldiscussionoftheH–KeffectandwewillbeginwiththePurkinjeeffect6,7andthenwewill discuss theHelmholtz color coordinates.8This will leadustothebrightness–,twosourceswith the same luminance but differentcolorsdonothavethesamebrightness).ThisisshownusingtheCIE1964chromaticitydiagram (from WyszeckiandStiles)9withloci ofconstantbrightnesstoluminanceratios Wenextsuper-imposeonthisplot,asetofthreeconstantpurity(satura-tion)lociof75%,50%,and25%,10showingthatthebright-nessincreaseswithincreasedpurity(saturation)andthatinmanyregionsontheCIEdiagram,thepurity(saturation)tthattheredLEDcanlookbrighterthanthegreenLEDisrein-forcedbytheworkofRonchi(brightnessmatchingofLEDs).11WenextdiscussDowling’spatentandtheKaiserequations,12which are mathematicallysimplercompared totheNayataniequations,13whichcanhelptopredictpartoftheH–owthisby examplesofmilitary workdoneonli,werelatethetheoryofcolorappearancemodelonperceptionandtheH–discusssomeaviationstudieswhereLEDground-landing-striplightslookbrighterthanfilteredincandescentlights.14ThisisfollowedbysomesimulateddashboardLEDlightingexperiments(usingLEDluminaires15ratherthanactualdashboardicons)presentsomeworkonAMOLEDcameradisplays16thatlookbrighterthanLCDs,followedbyLEDprojectiondisplays17thatlookbrighter thanconventionallampprojec-tion displays atthe same lastexample oftheH–Keffectistheobservationthatpico-onologicalreasons,weagainlookatthe2009patentbyDowling12onmixedLEDlightingwithcompensationfortheH–,weendwith somemorerecentstud-iesbeingperformedtopredicttheH–,wewillhaveshownthereaderamixoflaboratorywork,visualincidentaldiscover-ies,andvisiontheoriesoverthelast186years,allwhichshowthegrowingimportanceoftheH–horiswithDisplayDeviceConsultants,6170PlymouthRd.,AnnArbor,MI48105USA;telephone1+734/730-3116,e-mail:donofrio@.©Copyright2011SocietyforInformationDisplay1071-0922/11/1910-0658$1.00.658JournaloftheSID19/10,2011

2Purkinjeeffect/shiftPurkinje(aCzechphysician)7in1825discoveredthatredandbluecoloroveredthatpaintedredandblueroadmarkersshowedthatrelativebluepainttoredpaintbrightnessratioslookhigherwithadarkenedambient(atdusk)thantheydidwithahigherambientlevel(inthemorning).Anotherauthor6putsthedateat1823anddescribesthePurkinjeshift as theeffect thatmakesan object whichisredbydaylightappearblackbystarlightalowlightlevelsthesemarkersbecomeachro-matic(whichischaracteristicofscotopicvision).3HelmholtzcolorcoordinatesInthelate19thcenturyHelmholtz(1821–1894)nalsobeexpressedbysayingthattheH–Keffect“referstotheincreaseofbrightness(orlightness)forcolorhavingthesameluminancebuthigherchroma(orcolorful-ness)”(Ref.2,theMunsellsystem19:Chromaissimilartocolorsaturationorpurity).Munsellalsousltz definedhiscolor coordinatestobethedominantwavelength(λ(wherehighsaturationd)orfromRef.8(DW)andpurity(%P)impliesalowamountofwhitecon-tentforagivenhue).6Hence,(λatwo-dimensionald,%P)definesthecolorofalightsourcewithspacewhichcanbemappedintotheCIE(x,y)lengineersarefamiliarwiththeseparametersastheyaregenerallyusedtoevaluateLEDswhicharecommonlyseparatedintobatches(bins)basedontheir rangeof dominantwavelengths (seeFig. 1). Some me-tersusedincolormeasurementsinadditiontogivingthex,yoru′v′etheCIE1931x,ydiagramasseeninFig. 1, wesee the percentage of saturation or percent purity(%P)whichcanbeexpressedinthefollowingequation:%P=100×(S–N)/(DW–N)(Ref.8),(1)whereNisthe location ofthe white referencepoint,Sisthecolorpointbeingevaluated,andS–ointonthelocusandDW–,ycolorcoordinatesofthedominantwavelengthcalculatedfromthespectraldistri1923,Kohlrausch7discussedtheH–Keffectofcolorpurityandbrightnessandthoughtitwasveryimpor-tantinheterochromatic(color-matching)portanttorealizefromEq.(2)thatthevisualspectralluminousefficacyfunction(Vλ),whichistherelativesensi-FIGURE1—oftheeyeatvariouswavelengths(Ref.9,p.258)andthusafundamentalrelationshipinluminance(L)isdeter-minedbya “flickerphotometer”andnotasteady-statelight(Ref.9,p.262).L=kΎEλVλdλ(Ref.7),(2)whereListhephotopicLuminance,Eλrepresentstheenergydistributionofthestimulus,Vλistheunitaryphotopic-lumi-nousfunction,lyopticalscien-tistsknewthatcomparingvariouscolomanreportsthattheH–KeffectwasdiscussedinKohlrausch’s1935articleonColoredLightPhotometry.27However,fromLeGrand’sbook7onLight,Color,andVision, (suchasLeGrand)thoughtthatthateffectwasone ofsimultaneouscontrastsandcomplementarycolors,eyefatigue,1982editionofWyszeckiandStiles’(WS)bookoncolor9includesahistoryoftheH–Keffectfrom1952to1977,erochromicmatchingwasdoneusingabi-partitefield(Ref.9,p.414)io/TheHelmholtz−Kohlrauscheffect659

FIGURE2—Equi-brightnessratiosfordifferentcolorsamplesrelativetoawhitesample(W).4Brightness–luminancediscrepancyInFig.2,weseetheresultsofthestudyreportedbyWSwhichshow-brightnesslinesfromWS(Ref.9,p.414)tstimulus(heldat20cd/m2)usingabipartitefield,ultsofthesemeasurementsshowthatonemustraisetheluminanceofthewhitefield(asmuchas1.5timestheequi-luminancevalue)roundwasalsowhiteandheldat20cd/tudiesweredonewithilluminantCwhiteandshowtheexistenceofthebrightness–,even thoughthe twosourcesoflighthavethesameluminance,theymaynotbeperceivedtobeatthesamebrightness(sometimescalledlightness).Look-ingatFig.2fromaHelmholtzpointofview,ifweconsideraredlightsourcewithadominantwavelengthof610nmandifwemoveinastraightlinefromthecolorcoordinatesofthewhitefield(W)tothe610-nmpointonthelocusoftheCIEdiagram,theB/r,Fig.2alsoshows that ifthedominantwavelengthwasshiftedfromredtogreen(towards570nm),theB/pearsthatthereisbothawavelengthandapurity(saturation)dependencefortheB/h,superimposedonthepreviousfigure,wehaveaseriesofconcentriclocicorrespondingto75%,50%,and25%purity(saturation).660JournaloftheSID19/10,2011FIGURE3—1931CIEcolordiagramwithconstantB/Llinestowhichwehaveaddedpuritycontoursof25%,50%,and75%(Ref.21).Thisnewplot(Fig.3fromRef.10)showsanimportantconcept if the H–Keffectwasonlyduetopurityorspectral-line-widthnarrowing,assomeliteratureindicates,thentheiso-brightness-ratiolinesshouldconformtotheshapeofthepuritycontours,butinfacttheydonot!Theplotthusindi-catesthattheH–Keffectisnotjustafunctionofpuritybutisalsodependantonthedominantwavelength(whichisassociatedwithMunsell’sparameterhue).5BrightnessmatchesforLEDsIn1986,11Ronchi1986,sheusedanarrayofLEDs(redandgreen)raywasvieweddirectlybyanobserver. Alens beLEDshadcolorcoordinatesof(x0.309)withthereddominantwavelengthλr=0.690,yr=Rd= 620nmandthegreenLEDhadgreencolorcoordinatesof(xythegreendominantwavelengthg=0.408,g=0.589)withλkeptthe greenlight fixed andvariedthe luminancegd=565nm. kshowedthntionedbefore,thiseffectiscalledthebright-ness–tofthetwocolorsofthelampswereflashedonandoff(firstredandthengreenfor800mseceach).Whenusingdefocusedlight,shesaidthatevenlerimentalcon-ditionsusedawhitedaylightfluorescentambientlightwithaluminanceof1cd/m2.

TABLE1—Brightness–luminance results for threeLEDsusingtheKaiserequations.6KaisereffectequationsforcalculatingtheH–KKaiser,ina1986articleintheCIE,showedanempiricalrelationship[asdidSchrandainhisbookoncolorimetry5]forcalculatingtheH–quationswereusedbyDowling12(PhilipsSolid-StateLighting)inhis2009pat-entandseenasEqs.(3)–(5)atent,Dowlingusesluminancecompensationtocreatethe“vis-ual”colortemperatureofacolorlightmixture,subjecttotheconstraintsoftheH–er’smodel(whichisalsomentionedbySchandaasaCIEmodel),therearetwotermswhichmakeupbrightness;(1)acolorcomponentofbrightness(F)andtheluminancecomponentlog(Y)suchthatthebrightness(B)inKaiser’snotationL**isL**=log(Y)+F,(3)F=0.256–0.184y–2.527xy+4.65x3y+4.657xy4,(4)log(Y1)+F1=log(Y2)+F2.(5)FIGURE4—KaiserplotsoftheH-KeffectonaCIE1931x,ceivedrelativebrightness(RB)attheequiva-lentluminance(Y2=Y1)isRB=Y2/Y1=(10F1/10F2).(6)These equations12areusedbyKaisertocreateanewsetof iso-brightness plotsinthe1931CIEdiagramwhich at70Athrough70Farelociofconstantbrightness,thatis,bersonthecurvesinFig.4showthecalcu-latedbrightness–g.4,ablueLEDshouldlookbrighterthanaredLED,calculatedtherelativebrightness–luminanceratiosusingtheKaisermodelandthecolorsfortheblue,green,atashowthatatthesameluminancetheper-ceivedbrightnessoftheblue LED isabout55%higherthanthegreenLED,andthattheperceivedbrightnessoftheredLEDisabout37%higherthanthegreenLED.7ThePurkinjeeffectIn1991Ronchi21reportedherresultsonthePurkinjeeffectanddiscussedm-paringred and bluecolor caps (tiles) whichhavebeenset tohavethe sameluminance, thebluecap was initially te-worthytomentionherethatthelowestilluminanceusedinthisstudywasabout0.015cd/m2andittook20sectoob-servethetargetaet-upcondi-tionsaresomewhatsimilartothesimulateddashboardstud-iesweperformedin2007ondisablingglare15butRonchi’sworkwasdoneinthescotopticandmesopticeye-responseregions,llbediscussedlaterinthispaper.8NayatanimodeloftheH–KeffectIn1991,Nayatanietal.,intheirarticle“LightnessPercep-tionofChromaticObjectColors,”reportedtheirmathe-maticalmodeloftheH–nidiscussedtheH–Keffectagainin199713andobtainedindependentsub-jecttestingandhasgeneratedadifferentsetofiso-bright-nesscurvesandhasgedelhasbeenusedinio/TheHelmholtz−Kohlrauscheffect661

9Dim-lightaviationIn1996,Howard,23inhispaperonmilitaryaviationcockpitlightingandaperturelights,statedthat“ifwetakeasetofspectrallights,allhavingtheidenticalphotopicluminanceattheretina,ghtestcolors are the spectralblues,followedbythe spec-tralredsandthenthegreens.”Hispaperisadiscussionoftwo-anr-parametermethodusedatthattimeisthe“ColorAppearanceModel”(CAM)saysthat“theH–Keffectdependsonconeactivity;ittherefore plays a greater roleathigh mesopicandphotopiclightlevels.”WewillseesomeevidenceofthisinLao’spaper.1710Color-appearancemodels(CAM)In2005,Fairchild’sbookoncolor-appearancemodels25explained that theH–K effect pointsoutthe error inassum-ing thatthe Ytristimulusvalue (whichdefinesluminance)notcoverthecolor-appearancemodels,butonlysaythatmodelssuchasCIE-CAM97andCIECAM02arecoveredbyFairchildinhisbook.11LEDaviationsignalsIn2007,blue,green,edthathisworkbeganbecausetherewereanecdotalreportsthatrunwayLEDlampslookedbrightertgincandescentlightswithspecialfil-terstoapproximatethecolorsofthelampstotheFAAchromaticityboundaries,Bulloughfound thatthewhite ref-erencelampneededtobetwesthatweseetheH–Keffectbecausethethreeneuralchannelsinthebraincombinetoconstructperceptionsofsignal-lightbrightness:oneachromaticchannel(luminancechannel)diescon-firmed thatthe LEDlights didlookbrighterthanthe -pleteness,theseweremodelsbyWare,Guth,Kuehni,andNayatani.12DashboardlightlevelsIn2007,inaireswereusedtosimulatetheiconsusedonthedashboardsandinthecenterstack662JournaloftheSID19/10,2011TABLE2—constructedTable2toshowhowthemesopiclightlevelsareinvolvedwithautomo-tivespecificationsfordashboards,instrumentclusters,erallyacceptedluminancerangeformesopicvisionisbetween20cd/m2(20,000mcd/m2)and0.020cd/m2(20mcd/m2).Table2showsthusedthemcd/m2unitsbecausetheyarecommonlyusedinDepartmentofTransportationspecifications(Flor-idaandTexas).Theluminanceseenbythedriverfromroadsurfaces,laneindicators,xperiments,weshowedthatformesopicredandgreenLEDlightsourcesatequi-luminance,clearlyamanifestationoftheH–aalongwithour2008data25alsoshowthatatequi-lumi-nance,theredLEDluminairewill appear brighter thanthegreenLEDluminairefrom0.2downtoabout0.02cd/inancevaluesatandbelow0.02cd/m2,level,weprobablyhaveatransitiontotheeye’soisinagreementwithHoward’sworkwhichstatedthattheH–Kef2008resultsshowthatdis-ablingglareresponsetimeisrelatedtothebrightnessandnotluminance,,again,theredLEDmesopicluminairelookedbrighterandwasseensoonerthanthegreenluminaire.13AMOLEDdisplaymeasurementsIn2007,Tsujimura16inhisanalysisofAMOLEDdisplayssuggestedthattheH–KeffectisinvolvedwithAMOLEDdisplays(seeninFig.5).TheAMOLEDdisplaysthathave

FIGURE5—ashowedthatanAMOLEDdisplaywitha200-cd/m2lumi-nancelookedsomewhatbrighterthana300-cd/udywascomplicatedbythefactthatthelumi-nancedistatashowedthatthecolorgamut(seeninFig.6)idbefore,colorpurityisoneofthefactorsintheH–Keffect,whichimpliesthatthecolorswiththehigherpuritywouldlookbrighter,andthi,thisagreeswiththeobservationthattheAMOLEDdisplaywouldlookbrighterthantheLCD.14LEDprojectionTVIn2009,Liaopublished17anarticleonprojectionandH–servedthattheperceivedbrightnessoftheLEDprojectorwashigherthanthetraditionallampprojec-tor.+5Theircalculationswereforanadaptingluminancefrom10to10–2cd/6—A color-gamutcomparisonforLCDandOLEDdisplays.12FIGURE7—Brightnessratioasafunctionofluminance.12LiaoshowedthatbyusingNayatani’sequationsonecouldcalculatethattheLEDprojectorlookedbrighterthananultra-highpressure(UHP)por-tantisthatthecalculatedequivalentbrightness–luminance(Ltheeq/Y)ratio increased as theadaptingillumination loweredluminancetoabout20–30cd/atashow that the ratiooftheequivalentluminance/photopicluminanceL2.55,andtheLeq/YfortheredLEDis153/60oreq/YforthegreenLEDis173/165or1.05,thusindicatingforhistestthattheredLEDbrightnessisgreaterthanthegreenLEDbrightness,givenequi-luminanceconditions.15Pico-projectorsIn2009,Niesten18evaluatedthebrightnessoflaserpico-projectors. They said thattheuse oflaserscauses“saturatedcolorsinverynarrowwavelengths(smallerthanthatfromLEDs) tobe perceived 50% brighterthanthe broaderspec-trumoflightproduced byLEDs.”Their data showsthatthebrightnessofadisplaycanbeimprovedthroughtheuseoflasers.16CompensationmethodformixedlightingIn2009,Dowling12patentedamethodtocompensatefortheH–Keffectbyusinga“Look-UpTable”(basedonKai-ser’sequationspublishedin1986).Theseiso-brightnesscurvesareseeninFig.3andrepresentconstantB/rkmakesuseofKaiser’sequationsdiscussedinanearlierportionofthispaper.17RecentvisualmodelsIn a recentreview(internet) article Corneyet al.26fromtheInstitute19ofOphthalmology,London,UK,publishedin2009showshowmodelsofthehumanvisualsystemcouldbeconstructedtodemonstratetheH–io/TheHelmholtz−Kohlrauscheffect663

18SummaryWehaveshownthattheH–Keffectexplainswhylightsourceswiththesaer,wehaveshownthat both theimprovedspectralpurity(oranarrowerspectraldistribution)a,boththeHelm-holtzcolorcoordinatesof%Panddominantwavelength(whicharesimilartotheMunsellcolorcoordinatesofchromaandhue19)sshownbyWyszeckiandStilesintheir1931CIEdia-gramwithbrightness–luminanceplotsandoursuperposi-tionoflociwithreducingpurityfrom100to25%.Thisdatashowedthatthepurity(and/orcolorsaturation)owedthisdirectlywithcalculationsusingKaiser’edthatwhencomparingLEDswiththesamelumi-nance,theblueLEDisbrighterthantheredLED,mentalstudiesperformedbytheaxperimentshowedthattheH–Keffecthasbeenseeninmilitaryandcommercialaviationcockpits,groundlighting,aswellasinconsumerelectronicssuchasAMOLEDdisplaysandprojectors(bothLED andlasertypes).WehavealsomentionedsomeopticalstudiesconfirmingtheexistenceoftheH–Keffectandofferingtheorieswhythisoccurs(butthesearemorecom-plicatedthanwewanttopresenthere).ThispaperismeanttodescribetheautomotiveandlightingengineerinsightintothemanifestationsoftheH–KeffectandthelargerroleitwillhaveinLEDautomotivelighting,instrument-clusterlighting,centerconsoles,ildofRITforhisinformationontheH–ghetal.,“DiscomfortanddisabilityglarefromhalogenandHIDheadlightsystems,”AdvancedLightingTechnologyforVehicles(SPP-1668),SAE20002WorldCongress,Detroit,MI,Paper2002-01-0010(2002),p.1.2Dember&Warm,PsychologyofPerception,2ndEdn.(Holt,Rinehart,andWinstonPress,1979),io,“Anewapproachtodeterminingtheimageresponseofacolorpicturetube,”dcastandTelevisionReceiversBTR-19,No.3,145(1973).a,“Colorimetry,”OSA/AIPHandbookofAppliedPhotome-try,tis,ed.(1997),Chap.9,,“ReadabilityofColoredSignals,”CIESympo-sium,Budapest,Hungry(Sept.30,1999).n,HumanColorVision(Holt,Rinehardt,andWinston,1979),p.114.664JournaloftheSID19/10,d,Light,ColorandVision(ChapmanandHall,1968)(translation),p.113.8KonicaMinolta,“Thelanguageoflight,”p.11,/instruments/knowledge/light/pdf/language_of_Fig.1inpartfrom/Lab/Graphics/Col-ors/.9Wyszecki&Stiles,ColorScience,2ndEdn.(JohnWileyPublishers,1982),io,“A historical review oftheHelmholtz-Kohlrauscheffect,”SID 2009VehiclesandPhotonsSymposiumDigestofTechnicalPapers(2009).,“Successiveheterochromicbrightnessmatchesforan LEDdisplay,”ColorResearchandApplication11,No.2,164–168(1986).g,No7515128B2,Column5(KaiserEmpiricalFormula).ni,“SimpleestimationmethodsfortheHelmholtz-Kohlrauscheffect,”ColorResearchandApplication22,No.6,385–401(1997).ghetal.,“PerceivedbrightnessofincandescentandLEDaviationsignallights,”Aviation,SpaceandEnvironmentalMedicine78,No.9,893–900(2007).io,“BrightnessevaluationofLEDluminairesunderdynamicbackgroundchanges,”2007SIDVehicleDisplayandPhotons,Paper5.2(2007).uraietal.,“Advancementsandoutlookofhighperformanceactive/matrixOLEDdisplays,”SIDSymposiumDigest38(2007).al.,“PerceivedbrightnessofLEDprojector,”SIDSympo-siumDigest40,262–264(2009).netal.,“Picoprojectors:Nanosecondmodulationmakescell-phoneprojectorspossible,”LaserFocusWorld45,No.9,1–3(2009).l,AMeasuredColorSystem,BasedontheThreeQualitiesHue,Value,andChroma,2ndEdn.(o,Boston,1907).ald,ColorEngineering(JohnWileyandSonsLtd.),p.89,,“ThePurkinjeeffect:Someconsiderationsontheinterplayofreceptorpostreceptoralmechanisms,”ColorResearchandApplica-tion16,No.1,10–15(1991).22WiddelandPost,ColorinElectronicDisplays(PlenumPress,NewYork,1992),pp.21.23C. Howard,“Measuringvisual effectiveness ofdim light,”SIDSympo-siumDigest27,1–4(1996).ild,ColorAppearanceModels(JohnWileyandSons,Ltd.,2005),pp.119–io,“Astudyofmesopicadaptationconditionsfornightdriversusinganapproximationforoncomingheadlightglare,”eDisplaysandPhotons,84–88(2008).etal.,“Thebrightnessofcolor,”PMC,PLoSOne,2009;4(3)e5091(on-linepublication).man,ColorVision–AnHistoricalIntroduction(Wiley-In-terscience,1978),inappliedphysicsfromHofstraUniversityin1960,inphysicsfromNYUin1962, has been SIDRegionalVPo-author/heauthorofthechapteron“Displays”inventor/co-inventorin13patentsand11pesentlyas1996,heretiredfromPhilipsDisplayComponentsandstartedhisconsultingbusiness.