Albert Sadovnikov
COMPUTATIONAL EVALUATION OF PRINT UNEVENNESS ACCORDING TO HUMAN VISION
Acta Universitatis Lappeenrantaensis 412
Thesis for the degree of Doctor of Science (Technology) to be presented with due permission for public examination and criticism in the Auditorium of the Student Union House at Lappeenranta University of Technology, Lappeenranta, Finland on the 11th of December, 2010, at noon.
Albert Sadovnikov
COMPUTATIONAL EVALUATION OF PRINT UNEVENNESS ACCORDING TO HUMAN VISION
Acta Universitatis Lappeenrantaensis 412
Thesis for the degree of Doctor of Science (Technology) to be presented
with due permission for public examination and criticism in the Auditorium
of the Student Union House at Lappeenranta University of Technology,
Lappeenranta, Finland on the 11th of December, 2010, at noon.
MahineVisionandPatternReognitionLaboratory
DepartmentofInformationTehnology
FaultyofTehnologyManagement
LappeenrantaUniversityofTehnology
Finland
Reviewers ResearhDiretor,Doent,Dr. MarkkuHauta-Kasari
InFotonisCenter
UniversityofEasternFinland
Finland
ProfessorGöteNyman
DepartmentofPsyhology
UniversityofHelsinki
Finland
Opponents ProfessorJussiParkkinen
ComputerSiene
ShoolofComputing
UniversityofEasternFinland
Finland
ProfessorGöteNyman
DepartmentofPsyhology
UniversityofHelsinki
Finland
ISBN978-952-265-013-9
ISBN978-952-265-014-2(PDF)
ISSN1456-4491
Lappeenrannanteknillinen yliopisto
Digipaino2010
TheworkpresentedinthisthesishasbeenarriedoutattheLaboratoryofInformation
Proessingin theDepartmentofInformationTehnologyatLappeenrantaUniversityof
Tehnology,Finland,during theyears2003-2007.
For their nanial support I would like to thank the Papvision projet partners: the
European Union, Tekes, Stora Enso Oyj, UPM-Kymmene Oyj, Myllykoski Paper Oy,
MetsoPaperOyandLabvisionTehnologiesOy. Researhforthethesiswasonduted
under theTekesprojets70049/03,70056/04and40483/03.
Lappeenranta,August2010
Albert Sadovnikov
Albert Sadovnikov
ComputationalEvaluationof Print UnevennessAording to Human Vision
Lappeenranta,2010
95 p.
AtaUniversitatisLappeenrantaensis412
Diss. LappeenrantaUniversityofTehnology
ISBN 978-952-265-013-9
ISBN 978-952-265-014-2(PDF)
ISSN 1456-4491
Printquality andthe printability ofpaperareveryimportantattributeswhen modern
printingappliationsareonsidered. Inprintsontainingimages, highprintqualityisa
basirequirement. Toneunevennessandnon-uniformglossinessofprintedprodutsare
the most disturbing fators inuening overall print quality. These defets are aused
by non-idealinterations ofpaper,ink andprintingdevies in high-speedprintingpro-
esses. Sineprintqualityisapereptiveharateristi,themeasurementofunevenness
aordingtohumanvisionisasigniantproblem.
In this thesis,the mottlingphenomenon is studied. Mottling is aprintingdefet har-
aterized by a spotty, non-uniform appearane in solid printed areas. Print mottle is
usuallytheresultofuneveninklay-downornon-uniforminkabsorptionarossthepaper
surfae, espeially visible in mid-tone imagery orareasof uniformolor, suhas solids
andontinuous-tonesreenbuilds.
Byusingexisting knowledgeonvisualpereptionandknownmethodstoquantifyprint
tonevariation,anewmethodforprintunevennessevaluationisintrodued. Themethod
is ompared to previous results in the eld and is supported by psyhometri experi-
ments. Pilotstudiesaremadetoestimatetheeetofoptialpaperharateristisprior
to printing,ontheunevennessoftheprintedareaafterprinting. Instrumentalmethods
for print unevenness evaluation havebeenompared and theresultsof the omparison
indiate that the proposed method produes better results in terms of visual evalua-
tion orrespondene. The method has been suessfully implemented as an industrial
appliation andisprovedto beareliablesubstitutetovisualexpertise.
Keywords: qualityinspetion, paper industry, mottling, pereived unevenness, visual
pereption, psyhometris, automated optial inspetion, mahine vision,
image proessing and analysis, printing, print mottle, instrumental mea-
surements
UDC 004.932.2: 676.017.55
C50 yanat 50%density
C70 yanat 70%density
pd ylesperdegree
CSF ontrastsensitivityfuntion
HSWO heat-setweboset
HVS humanvisualsystem
HWC high weightoatedpapergrade
ISO InternationalOrganizationforStandardization
K50 blakat50%density
K70 blakat70%density
LGN lateralgeniulatenuleus
LWC lightweightoatedmehanialpaper
M70 magentaat70%density
MF mahinenishedpapergrade
MFC mahinenishoated
MFS unoatedmehanialspeialitypaper
MTF modulationtransferfuntion
MV mahinevision
MWC medium weightoated
RG rotogravureprinting
RGB red-green-blue
ROI regionofinterest
SC superalenderedpapergrade
SC-A superalenderedmehanialpaper
SFDA stohasti frequenydistribution analysis
SFO sheet fedoset
ULWC ultralightweightoated papergrade
WFC woodfreeoatedpaper
Y70 yellowat70%density
1 Introdution 11
1.1 Researhquestions . . . 12
1.2 Contributionsandpubliations . . . 13
1.3 Strutureofthethesis . . . 14
2 Paper and printquality 17 2.1 Paperanditsproperties . . . 18
2.2 Paperandprinting . . . 21
2.3 Printquality . . . 25
2.4 Printmottle . . . 26
2.5 Summary . . . 31
3 Human visionand psyhometris 32 3.1 Bakground . . . 32
3.2 Spatialvision . . . 35
3.2.1 Physiologialmehanisms . . . 35
3.2.2 Spatialfrequenytheory . . . 36
3.3 Contrastsensitivityfuntion. . . 37
3.4 Colorvision . . . 42
3.5 Psyhometri saling . . . 43
3.5.1 Diretintervalsaling . . . 44
3.5.2 Indiretintervalsaling . . . 45
3.6 Summary . . . 48
4 Models for mottlingevaluation 49 4.1 Overview . . . 49
4.2 Spatialdomain . . . 50
4.2.1 Speiperimeter . . . 50
4.2.2 ISO13660. . . 52
4.2.3 Stohasti frequenydistribution analysis(SFDA) . . . 53
4.2.4 Cluster . . . 54
4.3 Frequenydomain . . . 55
4.3.1 Band-pass. . . 56
4.3.2 ModiedBand-pass . . . 57
4.3.3 EnhanedBand-pass . . . 59
4.4 Colormethods . . . 60
4.4.1 Integrationmodel . . . 60
4.4.2 Pattern-olorseparablemethod . . . 62
4.5 Summary . . . 63
5 Experimentsand tests 65 5.1 Psyhometri experiments . . . 65
5.1.1 Diretsaling . . . 66
5.2.1 Summary . . . 74
5.3 Randomstimuligeneration . . . 75
5.3.1 Useoftheresults. . . 77
5.4 Preditingthemottle. . . 81
5.4.1 Papersamples . . . 81
5.4.2 Imagingandprintingsamples . . . 82
5.4.3 Experiments . . . 82
5.5 Summary . . . 86
6 Disussion 87
Bibliography 89
Introdution
Print quality is an essentialattribute when modern printing proesses are onsidered.
This is beause an inreasing proportion of data to be printed is in image form. If
the originalof aprint isassumed to be ideal,printquality depends on theprintability
of paper, printinginks, and the printing proess. Despite major improvements in the
aforementionedfatorsaetingthequality,thereareseveralundesiredeetsinprints.
One of the most severe defets is mottling, whih is the uneven appearane of asolid
printed area. It is related to the density and the gloss of print, and it is aused by
non-idealinterationsbetweenthepaperandtheinkinahigh-speedprintingproesses.
Inthesopeofthiswork,thetermsmottling,printmottleandprintunevennessreferto
thesameprintingdefet.
Reetive uniformity is what bothprinters and paper manufaturers striveto ahieve.
Whenthatgoalismissed,oneoftheresultingvisualappearanesmaybemottle. Mottle
is veryomplexinitsmanyausesanddiulttodesribefrom observationbeauseof
its manyvariations. Furthermore,thetermmottle isnotlimited touse intheprinting
and paperindustries. Furnituremanufaturersuse mottle to desribethe manygrains
of wood. Biologistshaveused mottleto desribethepatternsfound in butterywings.
Botanistsusemottletodesribeellstruturesin leaves. Mottleevenhasonedenition
for papermakersand another forprinters; however,the end usermakesno distintion
betweenthetwo.[5℄
Per-ÅkeJohanssondesribedmottleas:
...an unwanted laterally varying reetivity in homogeneous tone areas that expresses
itself asstohasti'loudiness'or'graininess'orsometimesin moreorderedpatterns. It
isnotvisibleintextareas,hardlyvisiblein'busy'imagesrihindetail,butmaybelearly
visible inalmerimageparts likeskiesorhomogeneousbakgroundtoneplates.[41℄
Mottling anbe dened asundesired unevennessin pereivedprintdensity. Figure1.1
givesavisual impression of thethesis topi. The InternationalOrganization for Stan-
dardizationhasalso denedprintunevennesusing twoterms,namelygraininess,whih
standsforaperiodiutuationsofdensityataspatialfrequenygreaterthan0.4yles
(a) (b)
(c) (d)
Figure 1.1: Samplemottling eetinprintpathes: (a)low mottling inblak
at50%;(b)highmottlinginblakat50%; ()lowmottlinginyanat50%;(d)
highmottlinginyanat50%.
per
mm
inalldiretions;andmottle,whihisdenedasaperiodiutuationsofdensityat aspatialfrequenyoflessthan0.4ylesper
mm
inalldiretions.[36℄Thisworkaimsatvisuallystohasti(aperiodi)mottleatfrequenieslessthan0.4yles
per
mm
, whih falls under the International Organization for Standardization (ISO) dened term.1.1 Researh questions
Thespei problemaddressedinthisthesisismottlingorprintunevenness. Thesetof
questions denedin thissopeareasfollows:
•
Whatarethepossibleausesofprintunevenness?•
Howdoesthehumanvisualsystempereivethemottlingphenomenon?•
Whatisthemodelofthehumanpereptionofmottling?•
Isitpossibletopreditmottlingpriorto printing?Several termsfrom the dened researhquestions needto belaried. Thevisual sys-
tem (humanvisualsystemin the sopeof this work) isthe partofthe entralnervous
system whih enables organisms to see, aswell as enablingseveral non-imageforming
photoresponsefuntions. Itinterpretsinformationfromvisiblelighttobuildarepresen-
tation of thesurrounding world. Thevisual systemaomplishesanumberof omplex
tasks, inluding thereeption of lightand theformation ofmonoularrepresentations;
the onstrution of a binoular pereption from a pair of two-dimensionalprojetions;
theidentiationandategorizationofvisualobjets;theassessmentofdistanestoand
betweenobjets;andtheguidaneofbodymovementsinrelationtovisualobjets. The
psyhologialmanifestationof visual information is known asvisual pereption, a lak
ofwhihisalledblindness.
A numberof methods for mottlingevaluation by an automatimahine vision system
havebeenproposed. TheISO13660standardinludesamethodformonohromeimages.
It isbasedonalulatingthestandarddeviationofsmalltileswithin asuientlylarge
area[36℄. Inthe standard,the sizeof thetilesisset toaxed value,whih isaknown
limitation[11℄. Therstimprovementtothestandardmethodwastousetilesofvariable
sizes [101℄. A number of methods relying on lustering, statistis, and wavelets have
also been proposed [5, 6, 93℄. Other approahes to evaluate gray-sale mottling have
their basis in frequeny-domain ltering [41℄ and frequeny analysis [79℄, whih were
thoroughlystudied in[83℄.
Themainontributionofthisthesisisthemethodforautomatedprintmottleevaluation.
A number of methods developed for the purpose of unevenness evaluation have been
previously published [5, 6, 41, 42, 40, 79, 101℄. Several of these methods laked the
orrespondene with human visual system (HVS) harateristis [36℄, other methods
weremissing anappropriatepsyhometri experimenttoderiveasuitableintervalsale
to measurethemottlingpereptionaurately.
1.2 Contributions and publiations
In thisthesis, themethod forthepereptionof printunevennessis derived andtested.
Severalresultshavebeenalreadypublished[86,83,84℄. Themethodhasbeensuessfully
implemented in industrial appliations and has proved to be areliable substitute to a
visualexpertise.
Beause thisworkonsistsofresultspreviouslypublished bytheauthor inanumberof
sientimedia,itisworthgivingashort summaryofthesepubliationshere:
In [47℄, the general approah to mahine vision appliations in the paper industry is
disussed. Theo-linepaperprintabilitytests,suhasheliotest,mottlingandpikingare
onsidered andtheinitialpossibilityofanautomatedinspetionisstudied. Theauthor
ontributesapreliminaryevaluation ofheliotestand mottlingevaluationmethods.
In[86℄, thegoalwastostudy themethodspresentedinliterature,and modifythem by
are disussedandompared basedontheirorrespondene tothehumanvisualsystem
and theexpertopinion of thegiven sampleset quality. The authorpartiipated in the
implementationofthemethods,statistisomputationandwritingthepubliation.
In [83℄, the analysis initiated in [86℄ is ontinued while the best performing bandpass
methodisenhanedtoovermorefrequeniesandtobetterorrespondto humaneval-
uation. The updated method is ompared to the ones disussed in [86℄. The paper
onludes that havingasingle,ontrastsensitivityfuntion (CSF) basedlterisbetter
thanasetofbandpasslters. Theauthordevelopedtheband-passmethodenhanement,
omputedperformanestatistisand partiipatedin writingthepubliation.
Publiation[85℄extendsthemethodintroduedin[83℄toolorprints. Anewexperiment
forolletingexpertknowledgeisperformed. Thesuggestedapproah,basedonpattern-
olor pereption separability,proved to orrelatewith thehumanevaluation well. The
authordevelopedthepsyhometriexperimentframeworkandpereptionmodel,ranthe
testsand wrotethepubliation.
In [49℄, the behaviorof gloss is studied using goniometri imaging and spetral olor
information. A goniometri imaging system is proposed to apture reetion images
that reveal thespatial variationof theglossoverpapersamples. Theharateristisof
thegloss aredisussedbyomparingtheresultsofgoniometriimaging,standardgloss
measurements and spetroradiometri measurements. The author partiipated in the
systemdesign,supervisedthetestsandreviewedthepaper.
Finally, [82℄ ombines the resultsof [85℄ and[49℄ to study the orrespondene between
pre-printarea images and gloss maps, and nal prints. The methods desribed in the
publiationusethegoniometriimagingsystemtoolletpre-printdataandpattern-olor
separable methodto omputemottlingindexesofnal prints. Theauthor partiipated
in experimentdesign,method implementationandwriting.
1.3 Struture of the thesis
Thespeiproblemoffastandreliablemottlingevaluationrequiressubstantialknowl-
edge inthefollowingelds:
•
papermaking,•
printing,•
imagequality,•
humanvisionandpereption,•
psyhometris,and•
mahinevision.Theproblemisbasedonvariousfatsfrom theaforementionedelds,butisnotlimited
to those. The thesistries to giveanoverview of allthe subjetsinvolved to provide a
solid bakgroundtothe ideasstated in theauthors'publiations. Figure 1.2showsthe
ChapterIIgivesaninsightintoprintqualityfromthepointofpapermakingandprinting.
Alsotherelationshipbetweengeneralimage qualityandprintqualityisdisussed.
Chapter III desribesrelevantfats from theelds ofhumanvisionand psyhometris.
Inthispart,thegroundsfordesigningexperimentsareestablished.
ChapterIVsummarizesknowledgeofdierentmottlingevaluationmethods. Itinludes
previouslypublishedapproahes,withsomehangesandupdatesthatareintroduedin
theaforementionedauthors'publiations.
Chapter V is about the experimental part of the work. Here, the psyhometris ex-
periments are desribed and a statistial analysis is given. This hapter also overs
goniometriimagingsystemdesignandthelatterresultsevaluation.
Finally,hapter VIpresentsashortsummaryanddisussionsetion.
Introduction ( Chapter I )
Discussion ( Chapter VI ) Print
Unevenness
Unevenness Evaluation
What is print unevenness?
( Chapter II )
What causes unevenness?
( Chapter II )
Which models are used for instrumental measurements?
( Chapter IV )
How the Human Visual System perceives unevenness.
( Chapter III )
How to extract knowledge about HVS unevenness perception.
( Chapter III )
How to set up an unevenness perception experiment.
( Chapter V )
Comparing instrumental and visual measurements;
evaluating results.
( Chapter V )
Figure 1.2: Contentsandstrutureofthethesis.
Paper and print quality
Paper making tehnology has a very long history ongoing from anient ivilizations.
During thelast 150years, it has greatlydeveloped and progressedsgniantly. In the
19thentury,Kellerdisoveredwoodpulp,andlatertheellulosetreatmentproesswas
invented,whihhasallowedtehnologytodevelopuptotoday'sstate. Currently,paper
is the major polygraphimaterial and meets alltehnologial, onsumerand eonomi
requirementsofsuhmaterials.
Tolaythe groundwork for this hapter, basidenitions should be given. First of all,
thetermpaperisgenerallydesribedasasheet materialmadeprimarilyofplantbers
[34℄. Note thatthe sope ofthis thesisislimited toaso alled graphi paper,in other
words,apapertypethat issuitableforprintingandwriting.
Theverbtoprintisdenedintheditionaryas"Transfertoasurfae;tomakeamarkon
asurfaebypressingsomethingontoit. OriginallyfromLatinpremere-topress." [72℄.
The sameditionary denes quality asadegree of exellene of something[72℄. Tobe
ableto onsiderprintquality, orin awidersense pereivedprintquality,it needstobe
lookedat asaombinationofthefollowingfators:
•
paperandinkproperties,•
printingproessissues,and•
pereivedimagequality.Asanbeseenfrom Figure2.1,thepereivedprintqualityisdependentonanumberof
fators,startingfrombasipaperandinkpropertiesandinludingmoreomplexaspets
ofprintingproess.
Perceived image quality
Printed image quality
Printing process quality
Paper properties
Ink quality
Paper quality
Ink properties
Figure 2.1: Simpliedshemaforpereivedimagequalitydependeneonbasi
paper/inkproperties.
2.1 Paper and its properties
Papermaking today inludes,in priniple,the sameproessstepsapplied for enturies:
preparation of the ber material, sheet or web forming, pressing, drying, sizing and
smoothing[34℄. Paperisanetworkformedbybresandbrefragments. Theproperties
ofalltheomponents,inludingllersandadditives,inuenethepropertiesofthepaper
andaetthequality. Singlebrepropertiesdesribestatistialpropertiesofbressuh
astheir length. Fibresan belassiedbygivenriteria,for example,in to earlywood
and latewood bres [54℄. Theanalysis of pulp bres is an importantareaof pulp and
paperqualityinspetion.
Paper harateristisvary widely, depending on thetypeof paper. Nevertheless,there
areanumberofpropertiesthatareharateristiofallpaper. Theseareinhomogeneity,
hygrosopiity,anisotropy,and visoelastisity[34℄.
Paper is an inhomogeneous material made from homogeneous elements: bers, llers,
and air-eld pores [34℄. Thehomogeneousregions'sizes vary from several mirometers
toseveralmillimeters(inthelongitudinalberdiretion),dependingontheharateristi
dimensionsofthesepartiles.
A sheet of paperalsoexhibits other inhomogeneities, whih anhave harateristidi-
mensionsoverafewmillimetersorentimeters. Theseareknownasloudinessandresult
from undesiredberoulationinthesheetformingproess. Inthisway,variationsin
aliperand density develop,andthis leadstoorrespondingvariationsin transpareny,
the"louds". Theprodutionproessanalsogiverisetoinhomogeneities,forexample,
duetostokpulsationsintheheadboxandinthesheetformingsetion[34℄. Onlybetara-
diographi measurementsangiveexatinformationaboutloal massdierenes. The
existene ofinhomogeneities must be takeninto aountin paper testing, forinstane,
byhoosingtheappropriatesizeof sample.
The most important harateristi of paper is its hygrosopiity [34℄, in other words,
its ability to absorb or release moisture, depending on the ambient limate, until an
equilibriumisreahed. Itissigniantwhetherthestateofequilibriumisestablishedby
absorptionor disorptionofwater.
As a resultof the hygrosopiity ofpaper, physial paper properties, suh assheet di-
stiness are dependent on the the ambient onditions. With a hange in the ambient
limate, thefullattainmentofanewequilibrium moistureontentofpaperisaproess
whih takesseveralhours. However,thehange in itsphysialpropertiesstartsalmost
immediately and ours veryquikly at the beginning and slowsdown later. Forthis
reason,papershouldbeproessedinroomsinwhihthelimationditionsarefavorable
forthepartiularpaperpropertyrequired.[15℄
Paperisananisotropimaterialwithregardtomanyphysialproperties. Thisanisotropy
is duetotheanisotropipropertiesoftheindividualbers,whih resultfrom thebril-
latedmirostrutureoftheberratherthanthebershape. Asaresultofthebrillated
struture, the ber an aept, for example, high tensile fores in the diretion of the
beraxiswithlowelongation;however,evensmalltensileforesatingperpendiularto
theberaxisausehighelongations.[34℄
The bers in a paper produed on a paper mahine are not aligned randomly: the
mahine diretion is usually preferred (anisotropy of ber orientation). Furthermore,
the ber matis passedthrough thepaper mahine under tension, whih prevents free
shrinkageofthebersduringdrying,mainlyinthemahinediretion.Restrainingfores
in therossmahinediretionarelowerandnonuniformarossthewidth, beingsmaller
at theedges. Thisresultsinanonuniformrossmahine proleofshrinkage.[34℄
Both ber alignment and shrinkage restraintsare responsible forthe anisotropy of the
moistureexpansionofnishedpaper,whihisgenerallyfarlowerinthemahinediretion
thanintherossmahinediretion,thelatterbeingnonuniformarossthewidth. Both
theabovefatorsalsoaettheload-deformationproperties(strain-to-stress)ofpaper,
whiharethereforealsoanisotropi.[34℄
Many papers exhibit an anisotropy with respet to their omposition in the normal
diretion (
z
-diretion). Theproesstehnologyof produtionon thefourdrinierwireis responsibleforthisphenomenon. Onthewireside,substanes(llersandnes)thatanpassthroughthewirearewashedout,while ontheotherside, theyare retainedbythe
bermat. Thisresultsin dierentsmoothnesspropertiesonthetwosidesof thepaper,
also knownas thetwo-sidednessofpaper. Two-sidednessofpaperanbeminimized by
twinwireformers. Afurtheranisotropyinthe
z
-diretionresultsfromthefrozenstressesduringnonsymmetrialdryingofthewebatthetopandbottomsides. Asaresultofthe
in-plain anisotropy ofthesheet, the mahine diretion mustbeonsidered andmarked
when takingsamplesfortesting anisotropiproperties,suhasstrength. Two-sidedness
is importantinthetestingof printability,for example. Thereforetopand bottom sides
havetobemarkedaswell.[34℄
Finally,paperhasvisoelastiproperties,inother words,itanbeelastilikeasolidor
visous likeathikliquid. The visoelastiityisalso aresultof thesuperimpositionof
thepropertiesoftheindividualbersandthoseofthebernetwork. Thevisousowof
theindividualbersisausedbytheslidingofbrilsathightensileloads. Theindividual
bers exhibit elastibehaviorat lowtensile loads. In theber struture ofpaper, the
sliding of bers at ber intersetions results in a ow eet and the paper undergoes
plasti deformation. A harateristifeature ofvisoelastiityis thedependene of the
onset ofowontheloadingrate. Smallloads atingoverlongperiodsof timeresultin
ow,whereaslargeintermittentlyatingloadsauseonlyelastideformation. Therefore,
tensilestressappliedausesthepapertobreak.[89℄
Thematerialpaperhasessentiallyfourharateristifeatures: inhomogeneity,hygrosop-
iity,anisotropy,andvisoelastiity[34℄. Thesefeaturesmustbetakenintoonsideration
in thetesting of paper. Themagnitude ofthese features depends onthe typeof ber,
theberrawmaterial, thepulpingproess,andontheproesstehniquesusedin stok
preparation and papermaking. Consequently, there are alarge number of variables in
papermakingandmanydegreesoffreedomareinvolvedintheadjustmentofthedesired
paperproperties.[34℄
An analysis of paperprodutsdenes the properties that are related to theuse of pa-
per. For ertain produts and produt behavior, dierent properties are required(see
Table2.1).
Table2.1:Requiredbehaviorofpaperprodutsandrelatedpaperproperties.[54 ℄
Required behaviorof paper Measurableproperties
Suientstrength Tensilestrength
Burstingstrength
Tearingstrength
z
-diretionalstrength Suitablestruture DensityAirpermeane
Optialproperties Brightness
Opaity
Color
Suitablesurfaeproperties Smoothnessorroughness
Surfaestrength
Surfaestiness Bendingstiness
Conoramediumtest
Edgerushtest
Manyphysial propertiesdesribepaperharateristis. Thepropertiesanbegrouped
asfollows[54℄:
•
basiproperties,•
strengthproperties,•
stinessproperties,•
struturalproperties,•
surfaeproperties,•
hygrosopityproperties, and•
optialproperties.Themostbasipropertiesofanypaperorboardinludemoistureontent,basisweight,
thikness,density,andlterontent. Paperandboardtradeisbasedonweight,therefore
basisweightlinksthepaperweighttoitssurfaearea. Thiknessanddensity,ontheother
hand, desribethe paperstruture. Paperstrength properties inlude tensilestrength,
bursting strength, internal tearingresistane, folding strength, surfaestrength and
z
-diretionalstrength [54℄.
Stiness relatesto amaterial's elastipropertiesand measures how muh thematerial
resists when it is deformed by an external load. Paper stiness is usually measured
astensilestinessand bending stiness. Tensilestiness ismeasuredby subjetingthe
papertoaforeparalleltothepapersurfae,andtheresultingdeformationiselongation.
Bendingstinessmeasuresthepaper'sabilitytowithstandabendingforewhenoneend
ofthepaperistieddownandaforeappliedto thefreeend[54℄.
Themostimportantsurfaepropertiesofpaperaresurfaestrength,roughnessorsmooth-
ness, fritionand gloss. Papersmoothnessorroughness desribespapersurfaetopog-
raphy. Papersmoothnessis obtainedbymeasuringairowbetweenapapersurfaeand
measuringsurfaeoredge. Informationneededforthesemeasuresarepressuredierene
usedtoreatetheairow,pressureofthemeasuringheadagainstthepapersurfae,and
the areaof the measuring head. The volume of air ow per time unit is reported as
roughnessandthetimeforaertainairvolumetostreamoutisalled smoothness. The
mostimportantpropertyintermsofthetopi ofthisthesisisgloss.[54℄
Glossmeasurestheinterationoflightandthepapersurfae. Therearefourbasiways
how light an interat with paper and usually, they an allhappen at the sametime.
Gloss measures thepaper'sabilityto speularlyreet light. High glossis desirablein
high quality paper with many images. Paper with high gloss has a wider tone range
than that ofamatte surfae. Thedownside ofhigh glossis that itusually impairs the
readabilityoftext andthereforein textbooksitis ahighlyundesirable property. Gloss
an be measured in many ways, but the paper industry adopted a 75
o
standard [94℄,
wherethemethodisintendedformeasuringthespeularglossofpaperat75
o
(15
o
from
the planeof paper). Although itsmain appliation is onoated papers,it is alsoused
foravarietyofunoatedpapers.
2.2 Paper and printing
Approximately 3000 dierent kinds of paper and board produts exist[69℄. Generally,
the produtswith aweightbelow225
g/m 2
isonsidered tobepaper,antthose aboveareonsidered tobeboard. Paperislassiedbasedonthefollowingriteria:
•
berfurnish ofthebase paper,•
oating,•
typeofsurfaenish,and•
enduse.Table2.2: Printingpapergrades. [69 ℄
Papergrade Desription
MF(mahinenish)News mehanialberbased,usedinnewspapersandpaper
baks
MFSpeial
SC(superalendered) unoatedwood-ontaininggravureandheat-setoset
paper;usedin magazinesandatalogs
MFC(mahinenishoated) oated (pigmented) wood-ontaining oset paper;
softalendered
LWC(lightweightoated)SC oated wood-ontaining gravure and heat-set oset
paper;superalenderedoline
LWCmatt mattalendering
MWC(mediumweightoated)SC doubleoated wood-ontainingosetpaper;
MWCmatt oatweightabovethelevelinLWC
HWC(highweightoated) doubleortripleoated wood-freeosetpaper;
HWCmatt oatweightabovethelevelinMWCgrades
WFC(woodfreeoated)
WFCmatt
basepaperisproduedfrombleahedhemialpulp,
withverylittleornomehanialpulp;
papersanbesingle,doubleortriple-oated;
thesurfaeistypiallyeithermattorgloss-alendered
unoatednepaper unoatedwood-freeosetpaper;
usedinbookprinting,opyingandeletroniprinting
ultralightweightwoodfreepaper unoated oroated wood-free"bible paper"; usedin
osetprintedbooks
The alenderingreferredto hereis anishingproess bywhih paper, plastis,rubber
andtextilesaresmoothed,glazed,polished,orgivenanembossedsurfae. Thematerial
is passed throughaseries of rollers, and the resultingsurfaedepends on thepressure
exerted bytherollers, their temperature, omposition,andsurfaedesigns, andon the
typeofoatingpreviouslyappliedtothematerialtobealendered.[34℄
Printingproessharateristisand itsresultingoutputdependnotonlyon paper,but
also on theinks used. Inksare dividedinto twolasses[81℄: printinginks and writing
inks. Printinginksarefurtherbrokendownintotwosublasses[81℄: inkforonventional
printing,inwhihamehanialplate omesinontatwithortransfersanimagetothe
paperorobjetbeingprintedon;andinkfordigitalnonimpatprinting,whihinludes
ink-jet andeletrophotographitehnologies.
Inindustrial statistis,printinginkmanufaturingislassiedasapartofthehemial
industry. Inkmakingsharesmanyfeatures withpaintmaking, but thereare alsosome
majordierenes. Onthelevelofpriniples,amajordierenearisesfromthefatthat
theolorofapaintedsurfaeisgeneratedwithonepaint,whereastheolorofaprinted
surfaeisgeneratedwiththreeormoreinks.[81℄
Atthemostgenerallevel,theompositionofprintinginksisguidedbythemainfuntion
ontrolled way. More speially, the hoie of raw materials and their quantitative
proportionsareinuenedbytherequirementsoftheprintingmethod andtheprodut,
inludingeologialrequirements.[81℄
The main omponents of printing inks are the pigment, the binder, and the arrier
phase[81℄. Inkmakingoperationsonsistprimarilyofhemialrawmaterialonversion,
andsurfaehemialandmehanialproessing.
Inkpigmentsaremainlysynthetiorganisubstanes,althoughsomemayontainmetal-
li elements. Blak pigments absorblight spetrally unseletively and olored inks se-
letively. Inkpigmentsshould not,when dispersed in the ontinuousphase of theink,
satter light[81℄.
A widerange of binders is used in inks. This is in partdue to the fat that available
binders originate from several soures inluding plants (drying oils), wood proessing
(pith, resins), rude oil distillation (hydroarbon resins), and hemial synthesis [81℄.
In part, this is due to variationsin therequirementsset onthe binders bythe arrier
phase of the ink and printing substrate. Thebinder and the arrierphase have to be
mutuallyompatible. Generalized, this meansthat the binderhasto besoluble in the
arrierphasein aontrolled manner. Thebinderandtheprintingsubstrateshouldalso
bemutually ompatible; thebindershould suientlyadhere and bind thepigmentto
the substrate. The overall quality level of printedmatter from a physial viewpointis
assoiated with the quality of the paper used. This being the ase the paper largely
determines the requirementson the binder. Asa rule of thumb, the binderontentin
inks is roughlythe sameasthe pigment ontent; typially"high quality" inks ontain
relativelymorebinder[81℄.
Aordingtothetypeofthearrierphase,inksaredividedinto thefollowinggroups:
•
liquidinks(gravure,exo),•
oilbasedinks(oset, letterpress,silksreen),and•
waterbasedinks(exo).Inoset,theombinationofthebinderandtheoilisalledavarnish,whereasthegeneral
termusedis"vehile". Thearrierphasebeomessuperuousaftertransferringtheink
to the paper and it is either removed by vaporation (gravure, exo, heat-set oset),
onvertedintosolidformbyoxidation(sheetfedoset(SFO))orpolymerization(SFO)
or it beomes absorbed into the paper (newspaperoset). In heatset oset, asilione
solutionisappliedafterdryingasathinlayer. Itfuntionsbyreduingthesurfaeenergy
oftheprintedlayer.[34℄
An inherentharateristiof mehanialprintingistheuseofthe impatprinipleand
mehanial pressure to transfer ink to the plate and from the plate to the paper. In
eletroni printing, ink transferto thelatent image, alled development, is aeted by
other typesof interativeforesbetweenthe inkand thelatent imagethan mehanial
fores. Theink'stransfertothepapermay,however,utilizemehanialfores.[34℄
In eletroni printing, no single-useplate for eah job is needed, and thelatent image
imageisgeneratedatall. Inthosemethodswherealatentimageisgenerated,itisbased
on interation of energy emanating from aprinting head with areeptor surfae. The
reeptor surfaemayalsobepaper.[81℄
Inthiswork,mottlingsampleswereprintedusingsheetfedoset(SFO)tehnology(see
Figure 2.2). SFO,heat-setweboset(HSWO) and rotogravureprinting(RG)printing
belongs to thelass of mehanialprinting. Mehanialprinting refersto themethods
of impat printing, whilst eletroni, often alled digital, printing refers to nonimpat
printingwhere datafromtheomputerisput ontothesheetofpaper[69℄.
Unprinted sheets
Inking system
Ink reservoirs
Cyan Magenta Yellow Black
Plate cylinder
Blanket cylinder
Impression roller
Dampening system
Figure 2.2: Sheetfedoset(SFO)pressshema.
Sensitivity to mottling is tested onoated papersusing HSWOand SFO printingand
bothsuperalenderedpapergrade(SC)andlightweightoatedmehanialpaper(LWC)
papersforRG. Sensitivityismeasuredinonnetionwiththepropertiesofpaperswhih
haveonnetionsto ink, oil and water interation. Mottling is measuredfrom printed
samples andink absorptionsamplesusing"wipe-out"tehnology. Coatedpapergrades
(woodfreeoated paper (WFC), medium weightoated (MWC), LWC, mahine nish
oated(MFC))forHSWO(SFO)andSCpapersandoatedpapers(WFC,MWC,LWC)
for RG are tested for mottling. Coatedgrades for HSWO (SFO) are both glossy and
matt,andonlyglossyforRG.
Figure2.3presentsageneralviewofanymehanialprintingmethod. Mehanialmeth-
odsin generalusetherotarypriniple,meaningthat theinkistransferredonthepaper
in a nip between tworotating ylinders. Ink transfersto the paper from theprinting
ylinder orfrom aseparatetransferylinder. Themethod thatusesaseparatetransfer
ylinderisalledanosetmethod. IntheRGmethod, anegativereliefisengravedinto
aopperlayerontheprintingylinder. Heat-setintheHSWOrefersto thesettingand
drying mehanismsin oset printing. Setting refers to a type of ink used in printing.
Heat-set inks ontainveryvolatileoils ompared to old-set inks. In theheat-set pro-
ess, theweb(rollsratherthantheseparatesheetsofpaperinSFO)isfed intoanoven
weight oils. A web typially travels throughfour printing units (see Figure 2.2), eah
printingadierent olor,and dryingours after these printingunits. RGprintingon
theother hand,integratesthedryinginto theunits,dryingtheink aftereaholorhas
beentransferredonto thepaper[69℄.
Inking unit Ink transportation and metering
Latent image on printing press
Ink application to printing plate
Data as ink distribution on plate
Ink transfer to paper
Drying
Printed paper Paper transportation
and registering Paper feeding
devices
Dryer
Paper Ink
Figure 2.3: Theimagingstepsinapress(toptobottom),supportingfuntions
(left),andrawmaterials(right)inprinting.[69 ℄
InFigure 2.3,theparametersthat anbealteredto ahievedierentprintqualitiesare
ink properties, ink feed, fount feed in the dampening proess, printing speed, drying
parameters,varyingrubberblankets,andprintingplates. Dierentlevelsofmottlingin
test printing an be reahed by hangingink transfer, watertransfer, ink setting, ink
absorption, water absorption, ink trapping, and drying parameters. Most hanges an
bemadebyadjusting theprintingproessparameters.
2.3 Print quality
Printedpaper in itsdierent manifestations isfamiliar to and usedby allpeople. The
timespenteahdaybrowsingprintedmatterforwork,eduation,informationandleisure
isonsiderable. Theneedstobefullled arediverse,asarealsotheusersituations.
Printprodutiontehnologieshavehangedimmenselyduringthelastfortyyears,more
sothanduringthepreedingfourhundredyears. Thisismainlyduetothedigitalization
of the basi rawmaterials of prints,namely information. Previousprogress was made
largely through mehanization. However, the appearane of printed matter has not
Fortraditionalgraphiprintedproduts,printabliltyisdenedasaombinationof:
•
runnabilityofthepress(lakofweb-breaks,aumulationproblems,fold-raking, blistering),and•
adequate print (produt) quality (olor reprodution, uniformity of print, print gloss,missingdots,printthrough).Despite all endeavourson the part of the paper industry, it is not possibleto reliably
predit either printability, runnability orprint quality. In other words, paper quality
annot be adequately haraterised as far asthese two aforementioned properties are
onerned.
Printability of paper is a paper property that indiates how paper behaves during a
printingproess. Printability depends on interations between paper and printing ink
and printingproessvariables. Good printability ofpapermeans that thepaperis not
very sensitive to variations in dierent proess variables and easily gives good print
quality.
Printquality desribes thenal resultof printing orthe quality of the printed image.
Printqualitydenition doesnot haveabsolute terms. It dependson theprintdensity,
resolution, and evennessof theprintedimage. Manyotherpropertiesare also ofinter-
est.[69℄
Printabilitytestsdependontheprintingproessinvolved. Certaintypesoftestmethods
aim at generalprintquality. These tests areusefulfor alltypesof papersregardlessof
the printing methods used. Some testmethods are only suitablefor aertainprinting
proess.
Printability and print quality depend on many dierent fators. The paper and its
propertiesareimportantomponentsasis theprintingproess withitsmanyvariables.
Due totheomplexnatureoftheprintingproess anddierentinterationsinuening
it, printabiliyandprintqualitytestinganonsistofdierenttehniques.
Papermakersandprintershavesearhedforafast,simple,andeetivetestmethodsto
predit paperbehaviorin printing withoutatuallyprintingthepaper. Everyprinting
proesshas its own features that often requireaseparate testing devie and a spei
test. In many ases, measuring similar properties after dierent proesses is possible.
Table 2.3 lists the laboratory printing tests ommonly used. Some tests may be used
with all the listed printingmethods, and some are more spei to a ertain printing
proess[69℄. AsitanbeseenfromTable2.3,testingformottlingisperformedonlyfor
HSWOand SFOprinting.
2.4 Print mottle
Printunevenness,oftenreferredtoasprintmottle,graininessorloudiness,or,generally
speaking, noise, redues the quality of the printed image. Mottling an be dened as
the unwantedunevennessin thepereivedprintolor,printdensityorprintgloss. The
seen in diuse but not speular illumination. If an image appears uneven in speular
illumination, it is due to gloss mottle. Topographi variations in the substrate, and
thereforealsoin theimage, analsobeareasonbehindmottledappearane.[41℄
Table 2.3: Laboratory test methods for prediting the printability and print
qualityindierentprintingmethods.[69 ℄
Oset Digital Rotogravure Flexo
Printing Printing Printing Printing
Coldset Heatset Sheet Ink-jet Laser
Printdensity x x x x x x x
Inkrequirement x x x x x
Inkset-o x x x x
Rub-o x x
Printthrough x x x
InkGloss x x x x x
Drypik x x
Wetpik x x
Blaktrappik x x
Mottle x x
Fiberroughening x x x
Toneradhesion x
Missingdots x
Residualsolvent x x
Dotgain x x x x x x
Dotsgeometry x x x x x
Unevennessan appearinmanyformsandatdierentsales. Typially,theunevenness
is stohasti, with lighter and darkerareas, with dierent sizes randomly spreadover
theprintedarea. Smallsale randomvariationissometimes alledgraininessand large
sale variation is alled mottling orloudiness. Aording to ISO 13360:2001[36℄, the
periodiutuationsofdensityat aspatialfrequenygreaterthan0.4ylesper
mm
inalldiretionsarealledgraininess(sparselyreferredtoasgranularity)andutuationsat
lowerfrequeniesare alled mottling. Sometimesthe unevenness inludes asystemati
omponent, when streaks, bands or other resular patterns are added to the random
unevenness[70℄. Typialprintunevennessdefets arepresentedinFigure2.4.
Thetypesofprintunevennessare:
•
one-dimensionalperiodinoise: banding(Figure2.4d),•
one-dimensionalrandomnoise: streaking(Figure2.4),and•
two-dimensionalrandomnoise: graininess/granularity(Figure2.4b),mottling(Fig- ure2.4a).Bandingisthepreseneofextraneouslinesonaprintedpage. Bandinggenerallyours
when aolor printer needs to pass the print headover apage multiple times to print
(a) (b) (c) (d)
Figure 2.4: Typesofprintunevenness: (a)mottle;(b)granularity;()streaks;
(d)banding.[93 ℄
Streaks are primarily one-dimensional visible defets in the image that run parallel to
theproessdiretion,alsoreferredto astheslow-sandiretion. Inauniformgraylevel
path, streaksmayappear asavariationin thetone level, whether the toneris blak,
yan,magenta,yellow,orsomeotherolor. Streaksinsingleolorseparationsthatmay
beunobjetionable,anauseanundesirablevisibleolorshiftforoverlaidolors.[63℄
There arethree main typesof mottling known in printing: absorption mottling, bak-
trapmottlingandwaterrepellanemottling. Thelastoneisseeninosetprintingwhere
thefountainsolutionwhihhasbeenemulsiedinto inkannot betransferred properly
on paper (paper is too dense or hydrophobiompared to the amount of fount fed in
printing). The original ause of mottling (absorption, baktrap or water repellane)
annot be deteted straightforwardfrom the printed samples - that an only be seen
whilewathingtheprintinganddoingsomehangesintheproess(inkfeed,fountfeed,
printingspeedet.) Thedierentmehanismsofmottlingourringindierentprinting
methodsforporousmaterials(notlms)aresummarizedin Table2.4.
Table 2.4: Mottlingourene mehanisms.
Printingmethod Mottlingmehanism
HeatsetWebOset(HSWO), Absorption,Baktrap
SheetFedOset(SFO) Waterrepellane,Piling,Contamination
ColdsetWebOset, Absorption,PrintThrough
Flexo,Rotogravure(RG) Piling,Contamination
Inkjet Absorption,PrintThrough
Paper formation and surfae harateristis ause the most basi set of mottle types.
Variations in thedensity, surfaesmoothness,berontent, nes, llerontent, sizing,
surfaepH,andspeienergyofthepaperanhaveadramatiinueneontheability
of anink to transferontothe surfae,itsabsorbeny,wiking, penetration,and drying
harateristis. Unevenberdistributionwithinthesheet(formation)isoftenattheroot
oftheseeetssinetheformationalsoinuenestheretentionofnesandadditives. A
poormassdistributionalsohasanegativeinueneonthelightsatteringoeientof
Mottling is usually the result of non-uniform ink absorption aross the paper surfae
and/oruneveninklay. Printmottleispartiularlyvisibleinmid-toneimagesofuniformly
oloredareas,suh assolidsand ontinuoustone sreenbuilds. Thisvisibleunevenness
anhappenasaresultofnon-uniformink gloss,densityorolorsofprintedinklm,or
it maybeaomplexfuntionof randomlyonnetedhalftonedots. Mottlingisusually
desribedmorespeially in termsof apartiularrootause. Insomesoures[5,24℄,
mottlingmehanismsareextendedinto awidersetoftypes,whiharelistedasfollows:
•
baktrapmottle,•
waterinterferenemottle,•
wet inktrapmottle,•
densitymottle,•
glossmottle,and•
drytrapmottle.Baktrapmottleisausedbytheinksinabilityto transferfromtheprintblanket tothe
paper. Typiallyaused whenthe ink is unstableorthesolvent/oil islost tooquikly,
anynon-uniformityin theink transferrateanreate amottledappearane. Baktrap
mottle usually appears in rst down inks, and is sometimes a result of improper ink
pH or visosity for the paper being used. Low visosity inks sometimes produe an
appearanealled'orangepeel',named forthepattern thattheinkreates[88℄. As the
sheet travelsfrom unit to unit,the ink lm non-uniformly traps bakonto subsequent
blankets resulting in uneven ink transfer to and absorption on the paper. The usual
reasons of bak trapmottling are paper uniformityand setting harateristis, ink set
rates,poorinktrap,andblankettype.
Waterinterferenemottleisthenon-uniformaeptaneorrejetionofthefountainso-
lutionarossthesheetsurfae. Typially,thismottlingmehanismisausedbyexessive
water, improperinkformulation,inadequately mixed solutions,orexessivealohol. If
thesheetdoesnotabsorbthefountainsolutionuniformly,theinkappliedinsubsequent
unitsmaynottransferandlayuniformly. Inkandwaterimbalaneanaettheunifor-
mityofinktransfer,resultinginanon-sharp,hollow,orweakdotstrutureinrespetive
units ofprint.[88℄
Ink trapmottle is aresult of aninorret ink takor pH sequene. Usually, aseond
downinkappearsmottledwhenitisprintedoverarstdownink,butappearsunmottled
whenprintedaloneonthesubstrate. Inktrapmottleisdesribedaspoororinonsistent
unit-to-unit ink trap whih transfers non-uniformly to the paper and/or previous ink
lms. Traprequiresonewetinklmtoaptureor"trap"subsequentinklms. Inorret
inktakgrading,wronginksequene,sreensoversolids,andpaperabsorbenyaremost
oftentheauseofinktrapmottling.[88℄
Density mottle is aused by uneven pressuresbetween metaltype, gravureylinder or
osetblanketandpaper. Itisnotommoninlithography,butanbeseenasaresultof
Glossmottleisanuneveninksurfaeharateristinotimmediatelyevidentatprinting.
It is aused by a paper's non-uniform absorption, resulting in portions of the ink to
appearglossyandothersmatte. Inkthatisimproperlyformulatedforapartiularpaper
isalsoaontributorto glossmottle.[98℄
Drytrapmottleisalsoalledrystallization: thisoursbeauseofanexessivedrytime
betweenruns ofamultiolorjob onasingleolorpressortheimproperwax ontentin
asetofink. Asinkdries,waxmigratestothetopsurfae. Exessivewaxmigrationtoa
dried inksurfaeanmaketheadhesionofsubsequentinksdiultbeausethesurfae
beomesslippery,hardandsmooth[98℄.
Therelationshipbetweenprintmottleandotherinhomogeneitiesinpapers,bothprinted
andunprinted,isworthtakingintoonsideration,andaredenedasfollows[27℄:
•
Mehanialproperties: Mehanialpropertiesofunprintedandprintedpaperrefer tothe strengthandstiness ofthepaper,andits apabilitytohandle stressesbystrainandompression.
•
Mehanialinhomogeneities: Mehanialinhomogeneitiesare spatialvariationsin mehanialproperties.•
Optialproperties: The paper'sand print'soptialpropertiesreferto theirmode ofinterationswithlight,inotherwords,howthelightisreeted,sattered,andabsorbedin thepaperorprint.
•
Optialinhomogeneities: Optialinhomogeneitiesarespatialvariationsoftheopti- alpropertiesintheprintedorunprintedpaper. Themostobviousformofoptialinhomogeneitystemsfrom variations in light absorptionovertheprintedsurfae.
Thisisseenasprintmottlein diuseordiretedilluminationorasglossvariation
inspeularviewing[59℄.
•
Printability: Theprintabilityofpaperis theombinationof paper-relatedfators thatontributetotheahievementofadesiredqualitylevel. Printabilityparame-tersaredividedintoasetofoptial,surfae,struturalandmehanialproperties,
whih are then measured. Some of the printability parameters refer to inhomo-
geneitiesin optialandmehanialproperties.
All of these properties and inhomogeneities, as well as their ombinations, aet the
resulting print mottle. Printingis an interation betweenthree main omponents: pa-
per, ink, and press. The auses of mottle anberelated to the propertiesof all three
omponents,ortotheinterationbetweenthem[27℄:
1. interationbetweenpressandink,
2. interationbetweenpressandpaper,
3. interationbetweenpaperandink,and
Mottlemehanismsvarydependingonthetypeofpaperandtheexisteneoftheoating.
In general, all printing methods using inks with solvents need mottling testing. The
main ausefor mottlein printingis theunevenabsorptionof inkand fountainsolution
(in oset) into paper,and eventhroughthepaper,ausing mottling. Other auses are
onneted to the wetting, trapping and setting of ink onto the surfaeof theprinting
substrate (normally paper). In oset printing, the ontamination of rubber blankets
and/orprintingplates ausemottledprintingaswell. This ontaminationantypially
beinkpilingonrubberblanketsoronprintingplatesineahprintingunit,pilingtherst
printed inks onto rubber blankets of latter printing units (arry-overpiling, bak-trap
piling) or ontamination on water areas of printing plates or rubber blankets in eah
printingunit.
Mottlingisnormallyevaluatedfromoverlappingoloursbetweenyan50%magenta50%
-yan70%magenta70%tones(C50M50-C70M70). Thereasonforthisistheproblem
onnetedtoredandblue shadesin printing(skin,sky). Tonesof50%strengtharethe
most sensitivefor pikingand ifitis theause ofontamination,this ouldbeseenas
mottledprinting. If pikingourss in allolors,it anbebestseenwith 50%tones in
eaholourofblak,yan,magentaandyellow. Absorptionandprintthroughmottling
of eaholour: yan,magenta,yellow,andblak(C,M,Y,K), isseenbest withsolidink
areas(100%)ofeahprimaryolor.
2.5 Summary
Inthishapter,thebasiaspetsofprintingandvariousonstituentsofthisproessare
onsidered. The main harateristisof thepaper, ink, and printmethods are dened.
Theommondefetsofsolidtone areareprodutionandtheironnetiontoparameters
oftheprintproessaredesribed. Themaindenitionsonnetedwiththeprintquality
arealsogiven.
The maintypesofprintmottle andthemehanismsof itsourenearepresentedand
desribed. Printingis aninteration betweenthree main omponents: paper, ink and
press. Paper,ink,andprintingproessproperties,andinhomogeneities,aswellastheir
ombinations,aettheresultingprintmottle.
Human vision and psyhometris
This hapter desribesthe human visual system and the aspets of the psyhometris
used in thisthesis. It startswiththephysiologialaspetsofthehumanvisualsystem,
ontinuingwiththespeisofthespatialandolorvision. Suhphenomenaashromati
adaptation,lightanddarkadaptation,andolor-patterseparabilityareonsidered. The
goalofthishapter istolaythegroundworkfortheexperimentalpartofthisthesis.
3.1 Bakground
In orderto understandthebasioneptsofhumanvision, thestrutureand thefun-
tionalityofthehumaneyeshouldbeonsideredrst. Figure3.1presentsarosssetion
of the human eye. The outer layerof the eyeis alled theornea. It is atransparent
tissue protetingthe eye, behind whih isthe aqueous humor, alear liquidlling the
avitybehindtheornea. Theorneaisaprimaryrefrativeinstrumentoftheeye,thus
serving astheforemostimageformingelementin thehumaneye[80℄.
The next part of the eye is the pupil, a dilating opening in the iris of the eye. The
iris expands orontrats, ontrolling the amount of light passinginto the eye [13℄. A
rystallinelensisloatedrightbehindtheiris. Theshapeofthelensisontrolledbythe
iliarymusles. Depending onthe shapeof thelens, apersonansee objetsloseror
fartheraway. Thehamberbehindthelensis lledwith learvitreoushumor,whihis
alearvisouswater-likeliquid. Finally,thebaksurfaeoftheeyeisalledtheretina.
Theretina has,infat,over100millionlight-sensitivephotoreeptorells,andit isthe
plae where thelightoming intothe eyeis onverted into neurohemialativitysent
intothebrain [100℄.
Photoreeptorsinthehumaneyeanbedividedintotwodistintlasses: rodsandones.
Theyare namedfor theirdistintshapes(seeFigure3.2). Eah ofthese photoreeptor
ellsperformsadierentfuntion. Notethat theamountof rods issigniantlyhigher
thanthatofones(approximately120millionvs. 8million). therods'primaryfuntion
is visionunder verylowlighting onditions,also alled sotopivision. Rods thus are
Figure3.1: Humaneyeinrosssetion[100 ℄.
extremelysensitivetolightandloatedthroughouttheretina. Conesontheotherhand
areonentratedintheenteroftheretinaandarenotassensitivetolightasrods. Cones
are responsible forpereption under thedaylight orsimilaronditions, alled photopi
vision. Theasewhenbothrodsandonesareemployedis alledmesopivision.
Figure3.2: Rodsandones[78 ℄.
Animportantfuntionoftheonesisthepereptionofolor. Thehighestonentration
of onesis loatedin the fovea,asmall regionin the enter oftheretina. Thisareaof
theretina aountsforthebestspatial andolorvision,thefovea, inturn, isproteted
byayellowlteralledthemaula,intendedtoprotetthefoveafromshortwavelength
Bothtypesofphotoreeptorshavesimilarstruture(seeFigure3.2). Theyonsistofan
outer segment, inner segment and a synaptiterminal. The outer segmentonsists of
billionsofpigmentmoleules sensitiveto light,whihin turnareproteinsalledopsins,
whihdeterminethewavelengthoflighttheyabsorb,andhromophores,apturinglight
protons. Eahphotoreeptoranontainoneofthefourpossibletypesofvisualpigments,
or opsins. Rods ontaina pigmentalled rhodopsin,while ones ontainthe pigments
responsible for responding to short, medium, and long light wavelengths. The inner
segmentofthephotoreeptorsonsistsofanuleusandotherellularmahinery[71℄.
Howanopsinatsingeneral,isitabsorbsaphotonoflightomingintoit,andtransmits
asignaltothephotoreeptorell,resultinginhyperpolarization. Figure3.3showshowa
humanretinaisonstruted. Asitanbeseen,photoreeptorssynapseintobipolarells,
whihthensynapseintotheganglionells. Theaxonsoftheganglionells,inturn,lead
to the opti hiasm viathe opti nerve. From the optihiasm there aretwoseparate
paths into thethe brain. The rstpart goes to the superior olliulus, that primarily
ontrolseyemovements. Theseond pathwayleadstolateralgeniulatenuleus(LGN)
andthenintotheprimaryvisualortex[71℄.
Light
Optic nerve fibers
Ganglion cells
Amacrine cells Inner synaptic layer
Horizontal cells
Receptor nuclei Bipolar cells
Receptors pigmented layer
Figure 3.3: Thehumanretina[71 ℄.
An important onept that needs to be onsidered here is the visual auity. What is
usually meantbythevisual auityisautenessorthedegreeof learnessofthevision,
whih depends on thesharpnessof theretinal fous. It anbeviewed asa measureof
thespatial resolutionof thehumanvisualsystem. Normalvisionisoftenreferredtoas
a20/20vision, whih if youput intonumbersmeansthat aperson withnormal vision
shouldbeabletodistinguish1arminuteatadistaneof6meters,andanresolvelines
with aspaingof1.75
mm
[17℄.3.2 Spatial vision
In this setion an integral view of human spatial vision is disussed, both from the
physiologialandpereptualpointsofview. Thevisualproessingandthefuntionality
of theprimary visual ortexare onsidered herein moredetail,aswellasthe ontrast
sensitivity funtion (CSF) of the human eye. In this hapter, CSFs for isohromati,
luminane-varying,one-dimensionalgratingsareonsidered.
3.2.1 Physiologialmehanisms
Asithasbeenpreviouslydisussed,retinalganglionellsonnettheretinaviatheopti
nerve to the brain. One of the most important harateristisof ganglion ellsis the
reeptive eld. What is meant by this term is the area in the retina that exites the
ganglionell. Itwasshownthatganglionellsrespondto spotsof lights,however,they
arealsoresponsibleforpereivingstripesoflightanddarkareasorsoalledgratings[52℄.
Ganglion ellsin theretinaareloatedinonentri irles. Partof theellsloatedin
airular areareattoan inreaseoflightomingin;these are so alledON-enteror
simplyONganglionells,theyaresurroundedbyellsthatgivethestrongestresponseto
thedereaseinlightintensity,thesoalled OFFells. Suhanorganizationofganglion
ells allows the retina to be sensitive to the dierene in the intensity of light in the
enterandtheperiphery,and berelativelyindierenttoaverageoveralllighting. Thus,
the retina atsas a lterthat inreases therepresentation ofontrast in thepereived
imageandltersoutambientlighting. itisassumedherethatontrastisadierenein
illuminationbetweenthemain objetandthebakgroundorbetweenlighteranddarker
areasintheobservedsene [100℄.
Figure3.4showsanONretinalganglionell'sresponsetogratingsofdierentfrequenies,
orinotherwordstodierentwidthsoflightanddarkareas. Whenthefrequenyistoo
lowor too high (see onsequentlya and), the response of the ganglion ellsis weak.
However,themedium frequenygratinggivesarelativelyhighresponse,leadingto the
onlusion that ganglion ellsare adapted to ertainspatial frequenies,meaning that
eah ellrespondsbest to a spei spatial frequeny, and as seenin Figure 3.4 worse
to higherandlowervalues[100℄. Anotheraspetof ganglionellsisthat theirresponse
depends onthe phaseofthe grating,whih isthe position within the reeptive eldof
theell[22℄.
TheaxonsoftheretinalganglionellssynapseontotwoLGN, whihatasarelayfrom
the retina to the primary visual ortex. They exhibit the same onentri positioning
asganglionells,andhavesimilarsensitivitiesto spotsoflightandgratingsofdierent
frequenies. LGN is, in turn, onneted to the erebralortex and other areas in the
brain thatontrolthefeedbaktothevisualinformationreeived[100℄.
TheareaoftheerebralortexthatreeivesdiretinputsfromtheLGNandisresponsible
forvisualinformationproessingisalledtheprimaryvisualortex,V1orstriateortex.
Theortexisdividedintotwohalves,allederebralhemispheres,that aresymmetrial.
The two sides are onneted via the orpus allosum, a large ber trat that allows
ommuniationbetweenthetwohemispheres. Thestriate ellshaveseveral important
(a)
(b)
(c)
Figure 3.4: Response of ganglion ell to gratings of dierent spatial frequen-
ies [71 ℄: (a) low frequeny yields weak response, (b) mediumfrequeny yields
strongresponseand()highfrequenyyieldsweakresponse.
frequenies. There are twoimportant featuresof V1 [29℄. The rstis the topographi
mapping,whihmeanstheimagefromtherighteyeismappedontotheleftstriateortex
and vie versa. The seond feature is alled ortial magniation. To put it simple,
it is the size of the ortial area dediated to a spei region of the visual eld, for
exampletheareadediatedtofovealvisionisgreaterthanthatoftheperipheralvision.
An importantonsequeneofortial magniationisthat visualauityissigniantly
reduedin proportiontothedistanetothefovea[53℄.
3.2.2 Spatial frequenytheory
Spatialfrequenytheoryisbaseduponaso-alledatomistiassumption,forexamplean
image ofanyomplexityanberepresentedwithprimitivespatialatoms. Intheframe-
workofthespatialfrequenytheory,thesearesinusoidalgratings. Sinusoidalgratingsare
two-dimensionalpatternsofwhihtheluminanevaries aordingtoasinewaveinone
ofthespatialdimensions,andremainsonstantintheother(perpendiular)dimension.
Eahofthesesinusoidalgratingsisompletelyharaterizedbythefollowingparameters:
spatial frequeny,orientation,amplitude,andphase[71℄.
Spatialfrequenysimplyisthewidthofthedarkand lightareasin thegrating,usually
expressed in the number of these yles per degree of visual angle. Orientation refers
Amplitude, often alled ontrast, in turn, an be desribed as the dierene between
the lightest andthedarkestareaswithin thegrating. Contrastisusually speiedasa
perentageof the maximal possible amplitudedierene, forexample 0% isequivalent
to auniform greyeld, while 100%is a blak andwhite eld. Finally, phase refersto
the position of the sinusoid relativeto aertain referene point. Phaseis speied in
degrees. Forinstane,a
0 ◦
sinusoidisalled asinephase,a90 ◦
,aosine phaseand, an180 ◦
,ananti-sinephase[71℄.3.3 Contrast sensitivity funtion
ACSFisonsideredtobeameasureofhowsensitiveanobserveristogratingsofdierent
frequenies. Note that in this hapter,CSFs forisohromati, luminane-varying,one-
dimensional gratings are onsidered. Typially, an imaging system in general, be it
a amera, or an optial system, is haraterized by the modulation transfer funtion
(MTF). In general, the modulation of a sinusoidal luminane pattern an be dened
as the amplitude of the sinusoidal grating divided by the average luminane [7℄. To
onstrut theMTF givento an inputof aertainamplitude, theoutputamplitude for
eahspatialfrequenyismeasured. Oneanmeasurethespatialresponseofasingleell
in thehumanvisualsysteminasimilarmanner. However,whendealingwithaomplex
system,suhasthehumanvisualsystem,reeivinganobjetive,aquantitativeresponse
isnotpossible[20℄.
Figure3.5: Frequenydependeneofontrastsensitivity,illustratedbythevary-
ing visibilityof atestpattern. Thefrequenyofthepatterninreasesgradually
fromlefttoright,andtheamplitudefromtoptobottom.[76 ℄
AsolutiontotheproblemofdeterminingtheshapeoftheCSFistorstndthelowest
detetable value ofontrast, whih isthe lowest valueof ontrastat whih anobserver
andetermine thedierenebetweenasinusoidalgratingand auniformgrey eld(see
Figure 3.5). Suh measurementsaredoneat variousspatial frequenies. CSFwouldbe
thereiproaloftheontrastthresholdrequiredforthedetetionof aertainfrequeny
grating. AnexampleoftheCSFshapeisgiveninFigure3.6. ThisisaCSFforanormal
human observer, obtained under photopi onditions, for a stati luminane varying
grating. Contrast sensitivity (y-axis) is found by taking the reiproal of the ontrast
threshold. Looking at the plot, itis lear that thehighest sensitivity is loatedin the
middle spatialfrequenylength,however,withtheinreaseof frequeny,thesensitivity
experienesasharpdrop.
and that whih falls above would non-visible. The inrease in ontrast sensitivity is
due to one spaing and the optis of the eye; the derease, however,is due to neural
fators[100℄.
Figure 3.6: Theontrastsensitivity funtion: windowofvisibility. Anyobjet
whosespatialfrequeniesandontrastsfallwithinthelightgrayregionwillbevis-
ible. Thoseoutsidethelightgrayregionareoutsidethewindowofvisibility.[100 ℄
Theeye'sontrastsensitivityvariesdepending onluminaneaswellasretinal loation.
Forhigh photopiluminane,thepeakofCSF liesathigherfrequenyvalues. Another
important harateristiof the shape of the CSF is that a sharp drop in the ontrast
sensitivityis exhibited foranyluminane, however,thedegreeof derease in thelower
frequenyrangedepends onthegeneralluminaneonditions[20℄.
CSF is often used in image quality researh. Suh measures as modulation transfer
area (MTFA) [16℄, integrated ontrast sensitivity (ICS) [96℄, subjetive quality fa-
tor (SQF) [31℄, disriminabledierene diagram (DDD) [14℄, and square-root integral
(SQRI)[9℄aredesribedin literature.
Contrastsensitivity funtion formulationsare derived using variousexperimental data,
the omplexity of whih depends on the tted data size and the intended generality.
There is anumberofformulationsavailable in literature[61, 99, 66,19, 7, 8℄. Someof
theformulationsarepresentedin Figure3.7.
One ofthe simplest ontrastsensitivity funtion formulationsis aparameterizedexpo-
nentialfuntion,rstdesribedinMovshonetal[66℄. Theformoftheequationisshown
in Equation3.1.
CSF (ω) = k s (ωk ω ) α e − βωk ω ,
(3.1)where
ω
isthespatialfrequeny. Thefourfreeparametersprimarilyaetthesteepnessof thelowfrequeny(
α
)andhigh frequenyβ
setionsoftheurve,lateralshiftsalongthefrequenyaxis(
k ω
),andvertialshiftsalongthesensitivityaxis(k s
). Thisfuntion10 -1 10 0 10 1 10 2 0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Movshon Mannos Daly Barten
Frequency [cpd]
S en si ti v it y
Figure 3.7: Dierentontrastsensitivityfuntions.
CSF (u) = au c e − bu ,
(3.2)where
u
isthespatialfrequeny,a
,b
,andc
arefree,adjustableparameters. Thisequation anbettedto existingluminane ontrastsensitivity dataifdesired. Forgeneraluse,the values of 75, 0.2, and 0.8 an used for
a
,b
, andc
respetively [43℄. This three parameterequationformsthegeneralband-passshapedesiredfortheahromationtrastsensitivityfuntion. Foruseinaolorimagedierenemetri,thetwo-dimensionalform
of the equation is used and the result is an isotropi funtion. The relatively simple
form of this model has both strengths and weaknesses. This funtion is the same for
all viewing onditions, unless the parameters are speially t to existing data. It
is generally assumed that viewing onditionsan greatly alter the ontrast sensitivity
funtion. This is espeially thease forluminane level,whih is known to atten the
shapeoftheontrastsensitivityfuntion. Tobetterpredithangesinviewingondition,
amoreompliatedfuntion isneessary.[43℄
Another relatively simple formulation (Mannos et al [61℄) also presents parametrized
exponentialandisgiveninEquation3.3.
CSF (u) = 2.6 (0.0192 + 0.114u) e − (0.114u) 1.1 .
(3.3)This funtion has apeak valueat afrequeny of8 ylesperdegree(pd) and azero-
frequeny interept of 0.5. A disadvantageof this CSF is that it does not aount for
theadaptivenatureofthevisualsystem[62℄.
DalydesribedanotherformofCSFforuseintheVisibleDierenePreditor[19℄. This
modelisafuntion ofmanyparameters,inludingradialspatialfrequeny,orientation,