Chapter 3 Measurement Procedures - NPSG.UW

1 Chapter3 Measurement ProceduresThegroupof measurements necessaryto characterizeboththecolorandsurfacefinish of anobjectis calledthemeasurementof appearanceof anobject[102].Thisgroupof measurements involves thespectralenergydistributionof propagatedlight, measuredin termsof reflectanceandtransmittance,andthespatia lenergydistributionof thatlight, measuredin termsof thespectraldistributionof thepropagatedlight affectappearancecharacteristicssuch as hue,lightnessandsaturation[102]. Hueis theattributeof colorperceptionby meanswhich anobjectis judgedtobe red,yellow, green,blue,purpleandso theattributeby which whiteobjectsaredistinguishedfromgray objectsandlight , saturationis theattributethatexpressesthedegreeof departurefromthegray of thespatialdistributionof thepropagatedlight affectappearancecharacteristicssuch asgloss,reflectionhaze,transmissionhaze, lusterandtranslucency.

2 Thereflectionhazecorrespondstothescatter ingof reflectedlight in directionsnearthatof specularreflectionby a specimenhavinga glossysurface[102].Thetransmissionhazeco rrespondstothescatteringoflight withinoratthesurfaceofanearlyclearspecim en,which is responsibleforcloudyappearanceseenby transmission[102]. Finally, theluster,or contrastgloss,as describedby HunterandHarold[102],correspondstotheglo ssassociatedwithcontrastsof bright andlessbright adjacent areasof thesurfaceof reflectedin thespeculardirectionandthatreflectedin thosediffusedirectionswhich areadjacent to MEASUREMENTPROCEDURES31 Greenberget al.[86]proposeda frameworktotest,validateandimprove thefidelity andefficiencyof computergraphicsalgorithms,whichis composedof threestages,namelylocallight scatteringmodels,lighttransportsimulatio nsandimagedisplay performingcomparisonsbetweensimulationsa ndactualmeasurements sothatsimulationscanbe usedin a predictive ,it is offundamentalimportancethatateach of reflectanceandtransmittanceareperformedu singspectrophotometers,andactualmeasurem ents of BRDFandBTDF areperformedusinggoniophotometers[102, 110].

3 Thesedevicesareimportantbasictoolsforfun damentalresearch in colorimetry[135],solarengineering[61],re motesensing[52, 108] andplant biochemistry[27,108]. Inthischapterwe discussthecomputersimulationsof such devices,henceforthcalledvirtualmeasureme nt to measurethespectraldatageneratedfromcompu termodelsandallowsusto performexperiments at different samplingresolutions,which areessentialrequirements forrenderingapplicationsas pointedoutby LalondeandFournier[126].Two virtualmeasurements aimedat , thesemodelscanbe verifiedby comparingtheirreadingswithactualmeasurem ents ,in orderto obtainthereadingsfromthecomputermodelsin thefirstplace,onemustperforma computersimulationof theinputsandoutputsof themodel, , consideringlight incident froma givendirectionandata givenwavelength,onemustmeasurethespectra landspatialdistributionsofthatlight afterbeingprocessedby areperformedusinga virtualdevice,or acomputersimulationof a realmeasurement device.

4 Whoseformulationshouldreproduceactualmea surement conditionsas faithfullyas possibleto minimizetheintroductionof biasin may be arguedthatwildlydifferent computermodelscanprovidethesamereflectan cefora givenilluminating,orincidence, ,forpracticalpurposestheevaluationof a computermodelwilltake into account how close,quantitativelyandqualitatively, theoverallCHAPTER3. MEASUREMENTPROCEDURES32curves providedby thismodelarefromtheactualcurves fordifferentmeasurement example,supposethatthespectralcurvesprov idedby a reflectancemodelA have anaveragediscrepancyof 5%withrespectto theactualcurves,andthecurves providedby a modelBhave anaveragediscrepancyof 30%.Which oneshouldbe incorporatedinto a renderingframework?

5 Thesecondapplicationcorrespondsto involve a largenumberofmeasurementswithrespecttodi fferent datacansometimesbe foundin theliteraturewhereactualmeasurements ,moreoftenit is notavailable,andevenwhenit is available,it is onlyfora restrictednumber of Measurement ,themostcomprehensive setof experiments involvingleafopticalpropertiesperformedt o date[101] was limitedto a singleangleof incidence,8 .Virtualmeasurement devicesareusuallydescribedin connectionwitha example,Gondeket al.[80] have useda deviceforspectralandspatialmeasurements, a virtualgoniospectrophotometer,presenteda sanopticsmodelanda capturedomeusedin conjunctionwitha geometricmodelof thischaptertheformulationof virtualmeasurement devicesis describedindependentlyof scatteringmodelsin thecomputergraphicsliteratureclassifieda s many casesnotentirelyaccuratesincetheyonlycom putetheBRDFandBTDF usingreflectanceandtransmittancevalues,w hich correspondto inputdata,as scalingfactorsor weights ,forthespatialdistributionof example,themodelsdescribedin Chapter6andChapter8 canbe usedto rendera plant leafunderdifferent lightingconditions, [93]

6 ,is,ingeneral, MEASUREMENTPROCEDURES33highlights two important issuesrelatedto ,it showstheneedfordevelopingmodelstocompute reflectancesandtransmittances, ,it showstheneedfordevelopingaccurateandeffi cient measurementprocedures,aspointedoutduring theWorkshoponMetrologyandModelingof ActualSpectrophotometersAspectrophotomet eris definedtobe any instrument formeasuringthespectraldistributionofref lectedandtransmittedradiant power,andspectrophotometryisdefinedasthe quantitative measurementofreflectionandtransmissionpr opertiesasa functionofwavelength[59].Spectrophotomet erscanalsobe usedto determinetheabsorptioncharacteristicsof anobjectas a functionof applicationsinvolvingorganicmaterials( , precisionfarmingandplant physiologystudies),however,themeasuremen t ofabsorptionprofilesisoftenperformeddire ctlywithfiber opticsmicroprobes [108].

7 Actualreflectancemeasurementsareusuallyp erformedunderilluminatingandviewingcondi tionsrecommendedby CIE2:45 /0 ,0 /45 ,diff/0 and0 /diff( ), reflectance(radiance)factorreadings, note,however,thatthenumericalvaluesof reflectanceandreflectancefactorareidenti calundertheconditionsof orviewingspecificationis total or diffuseonly .A glosstrapmay be incorporatedinthedesignoftheintegratings pheretoreducetheinfluenceof thespecularcomponent of CIE,in the45 /0 and0 /diffconditions,specimenswithsuch behaviorshouldnotbe measuredwithstrictlynormalilluminationin orderto reducethepossibility of introducingsystematicerrors,which may be causedby ,however,thattheanglebetweenthedirection of viewingandthenormalto thespecimenshouldnotexceed10 , andtheanglebetweentheaxisandany ray of anilluminatingbeam1 L MEASUREMENTPROCEDURES34shouldnotexceed5 [110].

8 0 /45 45 /0 0 /diff diff/0 integratingspheregloss :Typicalilluminatingandviewingmeasuremen t conditionsrecommendedby CIEforthecolorimetricspecificationof translucent specimensdependsgreatlyontheway specimensthetransmittanceis usuallymeasuredby placingthespecimenat theportof entranceof theinstrument[108] ( ).Therearea numberof detailedissuesspecifictoperformingspectr almeasurementsfortranslucent materialswhich arebeyondthescope of a comprehensive discussionof theseissuestheinterestedreaderis referredto thereportby AydinliandKaase[8].The0 /45 (45 /0 ) type spectrophotometershave beenmanufacturedwithspecimenareasupto approximately50mmin diameter,anda typicalsphere-type spectrophotometersexposesa smallerflatareaof specimen,roughly25mmin diameterformeasurement [102].

9 Anintegratingspheremay be of any diameterprovidedthatthetotalareaof theportsdoes notexceed10%of theinternalreflectingspherearea[110].Alt houghthereis nomaterialwithpropertiesof a perfectreflectingCHAPTER3. :Typicalmeasurement of ,it is possibletocalibratesuitable workingstandards ,suchas magnesiumoxide(MgO)powderor bariumsulfate(BaSO4) powder,which areoftenusedto cover absolutemeasurements,thespherewallis thestandard,andtheintegratingspheretheor y[76]compensatesfortheabsolutereflectanc eof thespherewallby mathematicallytreatingthewallreflectance as unity, sincethespectralreflectanceof such workingstandardsvariessomewhatwiththewav elength( [110]).Hence,thehemisphericalmeasurement smadewithsuch integratingspherescorrespondto absolutevaluesof reflectance(ortransmittance),which aresubjectto smallerrorsassociatedwithfactorssuch asaperturelosses,smallvaluesof non-uniformity of spherewallreflectanceandstray reflectancefromsamplemounts [212].

10 A typicalspectrophotometricrecordof reflectancemeasurements is a spectrophotometeris estimatedby theability oftheinstrument toreplicatea Measurement fora givenspecimenundersamespectralandgeometr icalconditions[110].Thebest-designed,bes t-constructed,andbest-calibratedspectrop hotometersstillyieldresultsfromthesamesp ecimenthatdifferfromonemeasurement to MacAdam[135],thedifferencesamongreadings shouldbe quitesmallandrandomlydifferent. Thesedifferences,or uncertainties,arenetresultsof combinationsof many smallfluctuationsdueto mutuallyunrelatedvariationsof different components of theinstrument, differentfactorsin theenvironment andhow thespecimenis theory,a spectrophotometeris consideredto be of highprecisionif thespectralmeasurements have anuncertainty, , of approximately [110, 135].