capturingtouch4

A.
Marcus(Ed.
):DUXU2014,PartII,LNCS8518,pp.
184–192,2014.
SpringerInternationalPublishingSwitzerland2014Scissors–APrecisePointingWidgetforTouchScreenDevicesFelipeBreyer,LuisA.
Vasconcelos,AntnioRivero,andJudithKelnerVirtualRealityandMultimediaResearchGroup,ComputerScienceCenter,FederalUniversityofPernambuco,Recife,Pernambuco,Brazil{fbb3,lalv,ahr,jk}@cin.
ufpe.
brAbstract.
Acommoncomplaintoftouchinteractionconcernsthelackofprecisionandfalsenegatives,especiallyinapplicationsinheritedfromthePCandmousecontext.
ThisworkintroducesScissors,avirtualwidgetfortasksthatrequireprecisionwhileinteractingwithtouchscreens.
Italsopreventsocclusionoftargets,iseffectiveforscreenedgesandiscompatiblewithcurrentinteractiontechniques.
Wedevelopedtwoprototypesfortesting:thefirstintroducedabasiclearningscenarioandthesecondpresentedtwosequencesoftargetswithdifferentsizesandlocations.
Thesystemrecorded,foreachtarget,thetimespentandnumberofattemptstoselectit.
Weperformedtestswiththirty-oneusersandevaluatedtheirdataaccordingtostatisticaltest,inthiscase,t-testfordifferenceofmeans.
TheresultsshowedthattheScissorswidgetwasveryeffectiveinthescenariosthatmotivateditsconceptionattaininganequivalentorupto11.
5timeshighersuccessrate,thusachievingitsmainpurpose.
Keywords:pointingwidget,interactiontechnique,touchdevices.
1IntroductionThepopularizationoftouchscreendevicesischangingthewayweinteractwithdigitalinformation.
TheNPDDisplaySearchestimatesthatsalesoftabletswillovercomenotebooksin2014anditwillbecomethemostusedcomputingplatform[5].
Accordingly,touchinteractionisrapidlyoutgrowingmiceandtrackpaddevices.
However,severaldesignchallengesfollowthischange(e.
g.
widerangeofapplicationswithdisplaysizes,resolution,andpixeldensity),andthefutureofthistechnologyrequiresdeepinvestigationinordertodesignabetterinteractionandmeetdifferentuserneeds.
Althoughsomepatternswerealreadyestablished[6]forthetouchscreensandinteractivesurfaces,thereisstillgroundforfurtherdevelopment;thevarietyofpossiblegesturesandwaystowhichsystemsrespondandinterpretcanopeninfinitepossibilitiesconsideringspecificscenarios.
Thispaperisorganizedasfollows;section3presentsrelatedworksandgivesdeeperperspectivearoundtouchinteractiontechniquesinordertostandasbasisforthedescriptionofthesolutioninsection4.
Evaluationmethodologyisdetailedwithtestprocedures,chosenusabilitymetricsandhardwaresetupsinsection5.
Section6Scissors–APrecisePointingWidgetforTouchScreenDevices185showstheresultsobtainedbyuserexperiment,withsomediscussionsonsection7andconclusionsandfutureworkspresentedinsection8.
2TouchDevicesTouchsensitivetechnologieshavebeenunderdevelopmentsincemid-'60s,however,theyonlyachievedmassivepopularityinthe2000s.
Touchdeviceshavesomeimportantcharacteristicsthatshouldbeoutlined,suchasitssizeandtrackingcapabilities.
Thesefeatureswilldefineforwhichtasksadeviceissuitable[2].
Inordertoanalyzeatouchdevice,firstly,weshoulddifferentiatethembetweentouch-tabletsandtouch-screens,thatis,ifthedeviceworksonlyforinput,astrack-pads,orifthereisatactile-displayworkingasinputandoutputsimultaneously,likepalm-tops.
Hereafter,anotherimportantcharacteristicthatdirectlyconcernsHCI(Human-ComputerInteraction)isthesizeofthedevice,whichwilldeterminewithhowmanyfingersorhandstheusermayinteractwithit.
Inturn,anotherfeatureandoneofthemostrelevantfortouchdevicesishowmanypointsitiscapableoftracking,i.
e.
single-touchormulti-touch.
[2]alsoremindsusthatifthedeviceissingle-touch,theinteractionparadigmwillnotchangewhencomparedwithmouse,trackballorjoystick.
Particularly,largeandmulti-pointtouchdevices,suchastouch-tablesandinteractiveboards,mayrequiretobeoperatedbymultipleusers.
So,thedevicecanutilizedifferenttechniquestodifferentiatethem,eitherbyindividualinteractionregionsoruniqueidsforeachstylus.
Thisbringsustoanothercharacteristic,iftheinteractionwillbedirectwiththefingersorwithsomeextratool,suchas,astylusoranothertangibleinterface.
Consideringtheinputinformationcapturedbythedevice,somearecapableofnotonlycapturingasinglepointatthetouchsurfacebutthewholecontactarea.
Thisway,wecandifferentiatetheinteractionthroughthetouchofafingerandwhenweholdthedeviceagainstourcheek.
Besidesthecontactarea,sometouchdevicescanalsoidentifytheangleofapproachandthepressureappliedtothesurface.
Lastly,onahigherlevelofabstraction,itisfoundthepossibilityofthesysteminterpretsgestures,requiringfromtheapplicationtotrackthepositionthroughoutthetime.
Differentinteractiondeviceshavevaryingcombinationsofthesefeatures,thus,thedesignermustbeawareofthemalltoadaptthesystemdesignandinteractiontechniquestobettersuitthehardwareinwhichtheapplicationwillrun.
3RelatedWorkAcommoncomplaintoftouchinteractionconcernsthelackofprecisionandfalsenegatives,especiallyinapplicationsinheritedfromthePCandmousecontext.
Todate,itispossibletofindcontextualsolutionsthatmagnifytheareabeentouchedandrequireusertoconfirmhis/herintention,commonlyseeninbrowsers.
Inothersolutions,suchasafewtextreaders,userscanzoomin,outandpan,repeatedly,thusavoidingfalsepositivesbutalsoreducingvisualization.
However,theseapproaches186F.
Breyeretal.
eitherpresentextraeffortfromusersorcauseinformationocclusion.
Occlusionmayalsooccurduetousers'fingersorhands,whileattemptingtoclickinabuttonorahyperlink,sincetouchscreensarebothinputandoutputdevices.
In[4],theauthorsconsiderintuitionasanimportantyetill-definedfactorwhendesigningeffectivemulti-touchinteractions.
Intheirresearch,theyperformanextensiveliteraturereviewonmulti-touchinteractions.
Basedontheirfindings,theresearchersconstructedaframeworkoffivefactorsthatdeterminetheintuitionofmulti-touchinteractions:direct-manipulation,physics,feedback,previousknowledge,andphysicalmotion.
Also,theauthorspointtwomajorresearchproblemsderivedfromtheincreasingcomplexityofmulti-touchapplications.
Thefirstisthatmulti-touchinteractionshouldnotbeconsideredinisolationbutaspartofawhole,andsecondly,evaluationsshouldbebroughtintoarealisticenvironment.
Theresearchin[1]dealswiththeintegrationofplanarmulti-touchsurfacesintothedesktopenvironmentinordertogiveusersanadditionalinputchanneltoenrichthecurrentinteractionparadigm.
TheauthorscreateaprototypecalledMagicDeskwhichenablesthemtotryvariouspotentialregionsandconfigurationsthatcouldbecomemulti-touchenabled.
Oneofthedrawbacksresidesinthenegativeseffectsassociatedwiththerightregionwhichwasdisplacedfromthekeyboardtoleaveroomforthemouse.
Thisfactreinforcestheneedfordevicesortechniquesthatallowspreciseinteraction,suchasmice,butnativetothemulti-touchparadigm,promotingarealcontinuousworkspace.
Thestudydevelopedin[3]presentsFluidDTMouseasasolutionformousesupportintouchapplications.
Theauthorsunderlinesomerequirementsinordertoemulateasmoothandnaturalmouseinteractiononatouch-basedsurface.
Firstly,itmustbeeasytopreciselypositionthemousecursor,whichisparticularlychallengingbecausetheuser'sfingeroccludesthecursor.
Also,itmustbesimpletotogglebetweenmouse-overandmouse-dragmodes,anditisundesirableforthistogglingmechanismtorequiremovementofthecursoritself.
Theirsystemdrawsthecursorbetweenthefirsttwofingerstotouchthescreen,andtheusermustactivatethecursorusingathirdfinger.
However,sincetheclickableareaisdefinedastheboundingboxcreatedbythetwofirstfingers,thisimplementationcausestheclickableareatodisappearwhenthefingersarealignedverticallyorhorizontally.
Anotherissueinthistechniqueconcernsthedifficultytoreachintocornersofthesurface.
4SolutionScissorsisavirtualselectingtoolfortouchscreendevices,itusesthreereferencepoints:twointerdependentfingersforpointingandoneforselecting.
Inordertopreventocclusion,theactivationpixelislocatedatthemidpointofthetwopointingfingers.
However,toselecttargetsonedgesorcorners,userscanapproximatetheirfingersand,whentheyreachathresholdof300pixels,Scissorsbendsintoanarrowheadshapecapableofpointingobjectsinsuchdifficulttoreachareas(Fig.
1A).
UsersmayalsoflipthedirectioninwhichScissorspoint,whichhelpsselectingtargetslocatedundertheirhands(Fig.
1B).
Thiswidgethasnocorrecthandpositionforitsuse:thetwoorientationpointscanbedefinedbyanycombinationofusers'Scissorsfingers.
Forinstance,usersfingersforpointing(Fig.
1othertouchinputwillactivaFig.
1.
ScissorscanbendindiffeAsrecommendedby[4],fivecharacteristics:direct-mandfeedback–Scissorswasavirtualwidget,theuserjoint/bendcapabilitiessimulookalikemetaphorevokestpreviousexperiences.
TheckeepstheuserawareofitscFs–APrecisePointingWidgetforTouchScreenDevicesswithlongnailsmightchoosetouseindexplusmidC),whilethumborringfingersforselecting.
Infact,atetheselection,asitisanindependentaction.
erentdirections(A,B)andusedwithanycombinationoffingers(awell-designedmulti-touchinteractionisdirectlyrelatedmanipulation,physics,physicalmotion,previousknowledsconceivedrespectingallthesefiverequirements.
Sincermanipulatesitdirectlyoverthesysteminterface.
Tulateaphysicsbehaviorandmotion,whilethescissthefeelingofaneverydayobject,favoringtheusertoreonstantfeedbackgivenbyScissorsduringitsdifferentstaconditionandinteractionpossibilities.
Fig.
2.
Scissorspossibleinteractions187ddleany(C)dtodge,it'sThesorsecallates188F.
Breyeretal.
Scissorswasdesignednottoconflictwithothercommontouchinteractiontechniquessuchaspinchtozoomandrotate.
Knowingthatpinchmovementisdoneovertheaxis(Fig.
2B)definedbyfirsttworeferencepoints,itwillnotactivateselection.
SincethedirectionthatScissorsbendsisdefinedbyamovementtothesamesideofthereferenceaxis(Fig.
2A),itdoesnotaffectrotationwhichisdonebymovingthefingersinoppositedirections,asshowninFig.
2C.
AnotherimportantfunctionintegratedwithScissorsisscrolling,positionedatScissors'referenceaxismiddlepoint(Fig.
2D);usersshoulduseindexfingerifusingthumb/middleasreferencesormiddlefingerincaseofindex/ringasreferences.
Finally,rightclickfunctionalitywasadaptedbydefiningaspecificregionatrightfromScissors(Fig.
2E).
Therefore,wewereabletomergeallfunctionestablishedformouseandtouchdevices.
5EvaluationMethodologyWedevelopedtwofunctionalprototypestoevaluatetheconcept.
BothprototypesweredevelopedwithlibGDX,agameandmultimediadevelopmentframeworkbasedonJavathatcanexporttheapplicationtodifferentplatformssuchasAndroid,iOSandHTML5.
Itfacilitatesthemanipulationandtrackingtouchesaswellasrenderingtasksfortheapplication.
TheprototypeswereinstalledonafirstgenerationGoogleNexus7tabletequippedwitha7-inchscreen,whichhas1280x800resolution,resultingintoascreendensityof216ppi.
ThefirstprototypeworkedasalearningscenariosettingusersfreetotryScissorssotheycouldgetusedtothisnewinteractiontool,asweconsideredunfairtocompareanewinteractiontechniquewithothertowhichuserswerealreadyfamiliar.
Thisprototypeworkedinaninfiniteloop,creatinghundredpixelssquarebuttonssotheusercouldtestScissorstoselectthematrandompositioninthescreen.
Theuserswereencouragedtotestthisprototypeaslongastheywanteduntiltheyfeltcomfortableandconfidentwithit.
Thesecondprototypepresentedtwosequencesofeighteensquaretargetsofsixdifferentsizesdividedinthreecategories:largetargetswith200and100pixels,mediumtargetswith50and25pixels,andthesmalleroneswith10and5pixels.
Thesetargetswerelocatedinthreedistinctscreenareas:center(ct),edges(ed),andcorners(cn).
Thetargetswouldreducetonextsizesaftereachthreedifferentregionswerecorrectlyselected.
Inthefirstsequence,weaskeduserstoaimandselecttargetsthewaytheytraditionallydointouchscreens,justtouchingthetargetswithafinger.
Inthesecondsequence,usersshoulduseScissorsforthesametask.
Asusabilityevaluationmetrics,thetimetocompleteataskwasusedasmeasureofefficiency,i.
e.
thetime(Tm)totouchorselecteachtarget.
Formeasuringeffectiveness,thenumberofattempts(At)totouchorselectthetargetbytheuserwasregistered.
Thesystemrecordedtheresultsforeachparticipantinalogfileforfurtherdataanalysis.
Scissors–APrecisePointingWidgetforTouchScreenDevices189Twohypothesesweredefinedfortheexperiment:H1:Scissorswillbemoreprecisethandirecttouchinteractiononsmalltargets.
H2:PrecisionprovidedbyScissorswillcompensatethetimespentwithitsmanipulation.
Weperformedthetestswiththirty-oneregularusersoftouchdevices.
16ofthemwereundergraduatestudentswhile15werepostgraduate.
AllsubjectswerestudentsattheInformaticsCentrefromFederalUniversityofPernambuco,Brazil.
Thegathereddatawereanalyzedaccordingtothet-testfordifferenceofmeans,inordertoevaluatethevariationbetweentheresults,aspresentedinnextsection.
6ResultsTheresultsshowedthattheScissorswidgetwasveryeffective,especiallyinthescenariosthatmotivateditsconception.
Theexperiment'sresultsaredeeplydiscussedbelow.
Ascanbeseenthroughthetaskexecutiontimechart(Fig.
3),subjectsusingScissorspresentedhighertaskdurationwhencomparedtotraditionaltouchinteractionforbigandmediumsizesinallconditions.
Scissors'worstresulthappenedintheeventof50pixelstargetlocatedonthescreencorner,obtaininganexecutiontimeapproximatelysixtimesslowerthantouchinteraction(Tm50cnTouch=0,8424s)(Tm50cnScissors=4,8936s).
Ontheotherhand,whenanalyzingthedatafromsmallertargets,10and5pixelslarge,itispossibletoobserveachangeinthisbehavior.
For10pixelstargetatthescreencentre,timewasconsideredstatisticallyequalforbothsamples(dTm10ct=0,0922>0,05),aswellasforthe5pixelstargetatscreencentreandcorner(dTm5ct=0,1520>0,05)(dTm5cn=0,8358>0,05).
However,itisimportanttonotethattheresultfortaskexecutiontimeatthescreenedgewitha5pixeltargetwithScissorswas50%lessthanwithtouchinteraction(Tm5edScissors=10,1243s;Tm5edTouch=18,8095s).
Fig.
3.
Meantimefordiversetargetsizesatdifferentlocationsofthescreen0510152025200ct200ed200cn100ct100ed100cn50ct50ed50cn25ct25ed25cn10ct10ed10cn5ct5ed5cnTouchScissors190F.
Breyeretal.
Onthenumberofattempts,mostoftheresultsforthelargerandmediumtargetswereconsideredstatisticallyequalbetweenusingScissorsortraditionaltouchinteraction(Fig.
04).
Theonlyexceptionsoccurredinthecasesofscreencentreandedgefor100pixelstargetwhichwereconsideredstatisticallydifferent(dAt100ct=0,0487<0,05)(dAt100ed=0,0100<0,05),evenwithnumericalvaluesclosetoasingleattempt.
Still,forthescreenedge25pixelstarget,Scissorsobtainedafavorableresultincomparisonwithtouchinteraction(dAt25ed=0,0446<0,05).
Withrespectto10and5pixelstargets,Scissorsshownbetterresultsinallthreeevaluatedregions,withhighlightstothecaseof5pixelstargetonscreenedge,inwhichScissorshasavalueeleventimesbetterthantouchinteraction(At5edTouch=38,9354)(At5edScissors=3,3548).
Fig.
4.
Userattemptsforselectingdiversetargetsizesatdifferentscreenlocations7DiscussionsBasedonthedataanalysispresentedabove,weconsiderthefirsthypothesisforthisresearch(H1)asconfirmed,duetothefactthatthenumberofuserattemptsfortargetselectionwithScissorswasconsiderablylower,confirmingitsefficiencyforsmalltargetsscenarios.
Thesecondhypothesis(H2)couldnotbeconfirmedbecauseScissorspresentsabetterexecutiontimeonlyforscreenedge5pixeltarget,whileinotherthreecaseswereconsideredstatisticallyequal(10ct,5ct,5cn),andforthetwoscenariosleft(10ed,10cn),theresultsfortraditionaltouchinteractionwerebetter.
Theseresultsrequireadeeperinvestigationandshowedthatmoretestsareneeded.
Moreover,somespontaneouscommentsfromparticipantswererecordedafterthetests.
Ingeneral,thevolunteersconsideredScissorseasytouse,claimingthatitspurposewasclearfromthemomenttheyrealisedtheneededtointeractwithgreaterprecisionwiththetouchscreen.
Frequently,theresearcherswerequestionedaboutitsofficialreleaseorwhereuserscoulddownloaditevenknowingitstillunder051015202530354045200ct200ed200cn100ct100ed100cn50ct50ed50cn25ct25ed25cn10ct10ed10cn5ct5ed5cnTouchScissorsScissors–APrecisePointingWidgetforTouchScreenDevices191development.
Particularly,womenwhoparticipatedintheexperimentshownsignificantinterestintheproject,duetothepossibilityofusinganytwofingerstomanipulateScissorsevenwithlargenails,whichfitstheirspecificneeds.
Onelastissuewecouldperceivewiththeexperimentisrelatedtosubjects'behaviortowardstheirfailure.
Duetothefrustrationofsuccessivelymissingsmallertargetsafterseveralattempts,aportionoftheusersstartedusingthestrategyoftappingthescreenrepeatedly,withoutcaringfortheconsecutivemistakes.
Inthescenariopresentedbytheprototype,therewerenootherinteractiveobjectssurroundingthetargetsothattheerrordidnotactivateanotherfunction.
Thisway,withoutthisriskoffalsepositives,itwasmorelikelythatusersadoptedthistappingstrategy.
Onthegathereddata,thisstrategycausedareductioninthetaskexecutiontimebutraisedthenumberofattemptsforsmalltargets.
Regardlesstheparticipant'sbehavior,traditionaltouchinteractiontechniquehasadisadvantageinoneofthemetrics,reaffirmingitsinadequacyincaseswhereprecisionisanimportantfactor.
8ConclusionsandFutureWorkInthisstudy,wepresentedthatmulti-touchinteractioncanstillbenefitfromappliedresearches.
Webelievetherearestillplentyofpossibledevelopmentsinthisfastgrowingtechnologyandithasnotreacheditsfullpotential.
Manyothermulti-touchapplicationsarestillunderdevelopmentfordifferentscenarioswithmanyspecificrequirementsandonlybyfocusingontheirspecialneedswemayachieverealinnovation.
Insuchmanner,theresultspresentedinthispaperencouragethecontinuityanddevelopmentofthisresearch.
Asthe'smalltargetscenario'canbefoundinmanymulti-touchapplicationsandsinceScissorsiscompatiblewithotherinteractionpatterns,webelievethisprojectshouldcoexistwiththecurrentinteractionparadigm,givingtheuserawiderrangeofchoices.
Accordingly,thisresearchfulfilleditsobjectiveasitpresentsaninteractiontechniquecapableofpreciseselectionasshownbythetestsresults.
Forfuturework,weintendtoperformexperimentsinsmallerdevicesinordertostresstheuseofthree-fingerinteractionandobservetheuserbehaviortogreaterocclusionsituations.
Itisalsoimportanttoinvestigatedifferentthresholdsfromthe300pixelsusedintheinitialdesignanddetermineitsimpactontheuserexperience;suchas200pixels,forexample.
Thecombinationoffactors,screensize,andthewidgetssizeshouldalsobeexamined.
Finally,sincetheinitialresultswerepromisingandweconsideredacontrolledenvironmentanadequatechoiceforafirstexperiment,arealtestscenarioshouldbeputtorun.
References1.
Bi,X.
,Grossman,T.
,Matejka,J.
:Fitzmaurice.
G.
:MagicDesk:BringingMulti-TouchSurfacesintoDesktopWork.
In:CHI2011,Vancouver,BC,Canada,May7-12(2011)2.
Buxton,B.
:Multi-TouchSystemsthatIHaveKnownandLoved(2013),http://www.
billbuxton.
com/multitouchOverview.
html192F.
Breyeretal.
3.
Esenther,A.
,Ryall,K.
:FluidDTMouse:BetterMouseSupportforTouch-BasedInteraction.
In:AVI2006,Venezia,Italy(2006)4.
Ingram,A.
,Wang,X.
,Ribarsky,W.
:TowardstheEstablishmentofaFrameworkforIntuitiveMulti-touchInteractionDesign.
In:AVI2012,CapriIsland,Italy,May21-25(2012)5.
NPDDisplaySearch:TabletPCMarketForecasttoSurpassNotebooksin2013.
In:QuarterlyMobilePCShipmentandForecastReport(2013),http://www.
displaysearch.
com/cps/rde/xchg/displaysearch/hs.
xsl/130107_tablet_pc_market_forecast_to_surpass_notebooks_in_2013.
asp6.
Saffer,D.
:DesigningGesturalInterfaces.
O'ReillyMedia,Sebastopol(2008)