Deconvolution is an indispensable tool in image processing and computer vision It commonly employs fast Fourier trans form FFT to simplify computation This operator however needs to t ransform from and to the frequency domain and loses spatial infor ID: 74437
Download Pdf The PPT/PDF document "Inverse Kernels for Fast Spatial Deconvo..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
InverseKernelsforFastSpatialDeconvolutionLiXuyXinTaozJiayaJiazyImage&VisualComputingLab,LenovoR&TzTheChineseUniversityofHongKongAbstract.Deconvolutionisanindispensabletoolinimageprocessingandcomputervision.ItcommonlyemploysfastFouriertransform(FFT)tosimplifycomputation.Thisoperator,however,needstotransformfromandtothefrequencydomainandlosesspatialinformationwhenprocessingirregularregions.Weproposeanecientspatialdeconvolu-tionmethodthatcanincorporatesparsepriorstosuppressnoiseandvisualartifacts.Itisbasedonestimatinginversekernelsthatarede-composedintoaseriesof1Dkernels.AnaugmentedLagrangianmethodisadopted,makinginversekernelbeestimatedonlyonceforeachop-timizationprocess.OurmethodisfullyparallelizableanditsspeediscomparabletoorevenfasterthanotherstrategiesemployingFFTs.Keywords:deconvolution,inversekernels,numericalanalysis,optimiza-tion1IntroductionDeconvolutionhasbeenanessentialtoolforsolvingmanyimage/videorestora-tionandcomputervisionproblems.Itwasalsousedinastronomyimaging[24],medicalimaging[9],signaldecoding,etc.Inrecentyears,itisextensivelyappliedtosystemsincomputationalphotographyandimage/videoediting,including\ruttershuttermotiondeblurring[19],generalmotiondeblurring[6,30,22,4,14,10,28,25,29,21],codedapertureanddepth[13,32],andimagesuper-resolution[2,23,17],sincemanytypesofdegradationcanbepartlymodeledorapproxi-matedbyconvolution,wherekernelsaremonotonicallydecayinglow-passlters.Whileconvolutioniseasytoapply,itsinverseproblemofproperlydeconvolv-ingimagesisnotthatsimple.Band-limitedconvolutionkernelshaveincompletecoverageinthefrequencydomain,whichmakesinversionill-conditioned,espe-ciallyundertheexistenceofunavoidablequantizationerrorsandcameranoise.Regularizationcanremedythisproblem{seeearlyworkofWienerltering[27]andTikhonovdeconvolution[26].Existingmethodsareintwostreams,whichhavetheirrespectivecharacteristics.SpatialDeconvolutionVeryfewdeconvolutionmethodsareperformedinthespatialdomain,owingtothehighcomputationalcost.Richardson-Lucymethod[20]doesnotinvolveregularizationandthusmaysuerfromthenoiseandring-ingproblems.Progressiveapproach[31]suppressesringingsbyoperationsinimagepyramids.Goodperformanceisyieldedinsparsepriordeconvolution[13], 2L.Xu,X.TaoandJ.Jiawhichrequirestosolvelargelinearsystems.Withthere-weightingnumericalscheme,thecoecientmatrixofthelinearsystemisno-longerToeplitzandcan-notbeacceleratedusingFFTs.Thisindicatesthatsparse-priordeconvolution,albeitusefulforpreservingstructuresandsuppressingringings,isnottranslationinvariant.DeconvolutioninFrequencyDomainTheconvolutiontheoremstatesthatspatialconvolutioncanbecomputedbypoint-wisemultiplicationinfrequencydomain,whichbringsoutpseudo-inversioninthefrequencydomain[16].Shanetal.[22]ttedthegradientdistributionusingtwoconvexfunctions.Thehalf-quadraticimplementation[11]mathematicallylinksgeneral-normstoafamilyofhyper-Laplaciandistributions.TheseiterativemethodsemployafewFFTsineachpass.EachFFTiswithcomplexityO(nlogn)wherenisthepixelnumberintheimage.Althoughfrequencydomaindeconvolutionisfast,itisnon-trivialforfurtherspeedupbyparallelization.Norisitsuitabletohandleirregularregions,whichhoweverarecommoninobjectmotionblur[3]andfocalblur[13].OurContributionInthispaper,weanalyzethemaindicultyofspatialdeconvolutionandproposeanewnumericalschemebasedoninversekernelstollthegapbetweenrecentfrequency-domainfastdeconvolutionandspatialpseudo-inverse.Theyareinherentlylinkedinoursystembyintroducingkernelsconstructedaccordingtoregularizedoptimization.Thenewrelationshipenablesempiricalstrategiestoinheritthenicepropertiesinthesetwostreamsofworkandtosignicantlyspeedupspatialdeconvolution.Althoughseveralusefulsparsegradientpriorsmaynotleadtotranslationin-variantprocessfordeconvolution.Wefounditispossibletoapproximatethemwithaseriesofoperatorsthatareindeedspatiallytranslationinvariant.Ac-cordingly,weproposeaneectivenumericalschemebasedontheaugmentedLagrangianmultipliers[15,1]andkerneldecomposition[18].Theresultingoper-ationsarenomorethanestimationofasetof1Dkernelsthatcanberepeatedlyappliedtoimagesiniterations.Unlikeallpreviousfastrobustdeconvolutiontechniques,ourmethodworksspatiallyandhasanumberofadvantages.1)Itiseasytoimplementandparal-lelize.2)ItrunscomparablywithorevenfasterthanFFT-baseddeconvolutionforhigh-resolutionimages.3)Thismethodcandealwitharbitrarilyirregularregionswithoutmuchcomputationoverhead.4)Visualartifactsaremuchre-duced.Weapplyourmethodtoapplicationsofextendeddepthofeld[12],motiondeblurring[29],andimageupscalingusingbackprojection[8].2MotivationandAnalysisTounderstandtheinherentdierencebetweenspatialandfrequencydomaindeconvolution,webeginwiththediscussionofconvolutionexpressedintheformy=xk+; InverseKernelsforFastSpatialDeconvolution3 (b)(c)(g)(a)(d)(e)()f()h Fig.1.Illustrationofregularizedinverselters.(f).(a)isaGaussianblurredimage.(b)-(d)aretherestoredimagesbyconvolvingtheregularizedinverselter,Wienerdeconvolution,and1DseparatedWienerdeconvolution.(e)showstheGaussiankernel.(f)showsthedirectinverselter,andregularizedinverselterfromtopdown.(g)contains1Dscanlinesofthetwoinverseltersin(f).(h)showstheclose-upsof(c)and(d).wherekisthekernel,yisthedegradedobservation,xisthelatentimage,referstotheconvolutionoperator,andindicatesadditivenoise.WerstexplaintheinversekernelproblemusingthesimpleWienerdecon-volutionandthendiscusstheissuesindesigningapracticalspatialsolverusingsparsegradientpriors,whichiseectivetosuppressnoiseandvisualartifacts.2.1SpatialInverseKernelsforWienerDeconvolutionWienerdeconvolutionintroducesapseudo-inverselterinfrequencydomain,expressedasW= F(k) F(k)2+1 SNR(1)whereF()denotesFouriertransformand F()isitscomplexconjugate.SNRrepresentsthesignaltonoiseratiothathelpssuppressthehighfrequencypartoftheinverselter.Therestoredimageisthusx=F 1(WF(y))(2)whereF 1istheinverseFouriertransform. 4L.Xu,X.TaoandJ.JiaAlbeitecient,restorationusingFFTslosesthespatialinformationasdis-cussedaboveandcouldbelessfavoredinseveralapplications.Thismotivatesustoapproximatethisprocessusingpseudo-inversewinthespatialdomain,expressedasx=F 1(W)y=wy;(3)wherewisthelatent(pseudo)spatialinversekernel.Itisknowninsignalpro-cessingthatthistaskcannotalwaysbeaccomplishedgivenanarbitraryWTakingthesimple2DGaussianlterforexample(Fig.1(e)),itsdirectspatialinversekernelisa2Dinniteimpulseresponse(IIR)lter,asshowninthetopofFig.1(f).Contrarily,wefoundthatthespatialcounterpartofWienerinversion,i.e.F 1(W),hasanitesupport,asshowninthebottomofFig.1(f).Thedier-enceisduetotheinvolvementofregularization1=SNR.Itisactuallyageneralobservationthatinverselterswithregularizationaretypicallywithdecayingspatialresponses.An1DvisualizationisgiveninFig.1(g).Thekernelwithregularization(bottom)decaysquicklyandthushasacompactsupport.AnimagedegradedbyaGaussiankernel(Fig.1(e))isshowninFig.1(a).TherestoredimageusingthespatialinversekernelwithcompactsupportisgiveninFig.1(d),withvisualartifactsnearimageborder,whichcanbeamelioratedbypadding.Tofurtherincreasethesharpnessandsuppressartifacts,weturntoamoreadvancedsparsegradientregularization.2.2SparseGradientRegularizedDeconvolutionState-of-the-artdeconvolutionmakesuseofsparsegradientpriors[13,11],mak-ingtheoverallcomputationmorecomplexthanaWienerone.Inthispaper,weproposeapracticalschemetoachievespatialdeconvolutionevenwiththesechal-lenginghighlynon-convexsparsepriors.Wedescribetwoissuesinthisprocess,whichconcernkernelsizeandnon-separabilityofregularizeddeconvolution.KernelSizeSpatialinversekernelscouldbeofconsiderablesizes.ForaGaussiankernelwithvariance=3,thecorrespondingregularizedinverselterusingEq.(1)hasanitesupportof5151.Althoughitisindependentoftheinputimagesize,itstilllaysalargecomputationalburdento2Dconvolution.KernelNon-separabilityManykernelsareinherentlynon-separable.Evenforthosethatareseparable,theirinversionsarenot.Forexample,eachGaussiankernelcanbedecomposedintotwo1Dlters,appliedinthehorizontalandverticaldirectionsrespectively.However,itsinversionisnotseparableduetoregularization.Theroadtospeedingupregularizeddeconvolutionbysimplyperforming1Dlteringisthusblocked.ThecomparisoninFig.1(c)and(d)illustratesthedierence.Thereisa2DinversekernelofGaussiancreatedaccordingtoEq.(1)andaseparatedapproximationusingouterproductoftwo1Dlters,formedalsofollowingEq. InverseKernelsforFastSpatialDeconvolution5(1).Therestorationresultusingthere-combined1Dltersisshownin(d).Itcontainsobviousoblique-lineartifacts(seetheclose-upsin(h)).WeaddressthesetwoissuesusingkerneldecompositionwithSVD,presentedbelow.3SparsePriorRobustSpatialDeconvolutionSparsegradientregularizeddeconvolutionworksverywellwithahyperLaplacianprior[11].ItminimizesthefunctionofE(x)=nXi=1 2(xk y)2i+c1xi+c2xi(4)whereiindexesimagepixels.c1andc2arenitedierentialkernelsinhorizontalandverticaldirectionstoapproximatetherst-orderderivatives.controlstheshapeofthepriorwith051.AcommonwaytosolvethisfunctionistoemployapenaltydecompositionE(xz1;z2)=nXi=10@ 2(xk y)2i+Xj2f1;2 2(zj cjx)2i+zji1A(5)wherez1andz2areauxiliaryvariablestoapproximateregularizers.Theproblemapproachestheoriginaloneonlyifislargeenough.Thesolveristhusformedasiterativelyupdatingvariablesaszt+1j argminzE(xt;zj;t)(6)xt+1 argminxE(x;zt+1j;t)(7)t+1 2t(8)tindexesiterations.Sincezjhasananalyticalsolution(orcanbefoundinlook-uptables)[11],themaincomputationliesintheFFTinversionsteptocomputex,whichgivesx=F 1 Pj F(cj)F(zj)+ F(k)F(y) PjF(cj)2+ F(k)2!(9)ItinvolvesseveralFFTs.Basically,updateofzjisperformedinspatialdomainasitinvolvespixel-wiseoperations.Sodomainswitchisunavoidable.3.1PenaltyDecompositionInverseKernelsWeexpandEq.(9)bydecomposingthenumeratoranddenominatorandapplyinverseFFTseparately.Ityieldsx=F 1 1 PjF(cj)2+ F(k)2!0@Xjc0jzj+ k0y1A(10) 6L.Xu,X.TaoandJ.Jiawherec0jandk0areadjointkernelsofcjandkbyrotatingthesekernelsby180degree,andjindexesdierentialkernelsc.Theoperationsc0jzjandk0yarenowinspatialdomain.k0yisaconstantindependentofvariableszandx(PF(cj)2+ F(k)2) 1inEq.(10)istheinversioninthefrequencydo-main.Itsdomainswitchtopixelvalues,infact,correspondstoaspatialinversekernel.Theregularizationmakesitsnitesupportexist.Soitispossibletoestimatespatialinversekernelscorrespondingtothisterm,i.e.,w=F 1 1 PjF(cj)2+ F(k)2!(11)Thisprocessraisesatechnicalchallenge.Becausevariesiniterations,wneedstobere-estimatedineachpass.Aseriesofspatialinverseltersthusshouldbeproduced,whicharenotoptimalandwastemuchtime.3.2AugmentedLagrangianInverseKernelsTotthespatialprocessingframework,weadopttheaugmentedLagrangian(AL)method[15,5]toapproximatedeconvolution.ALwasoriginallyusedtotransformconstrainedoptimizationtoanunconstrainedonewiththeconven-tionalLagrangianandanadditionalaugmentedpenaltyterm.Specically,wetransformEq.(4)intoE(xzj;\rj)=nXi=10@ 2(xk y)2i+Xj2f1;2zji+Xj2f1;2 2(zj cjx)2i h\rj(zj cjx)ii1A(12)wheretheterminthesecondrowistheaugmentedLagrangianmultiplierspecicforthisproblem.hiistheinnerproductoftwovectors.Themajordierencefromtheoriginalpenaltydecompositionoptimizationisthatheretheupdateof\rjpreventsfromvaryingwhiletheoptimizationstillproceedsnicely.Theiterativesolverisgivenbyzt+1j argminzE(xt;zj;\rtj)(13)xt+1 argminxE(x;zt+1j;\rtj)(14)\rt+1j \rtj (zt+1j cjxt+1)(15)Fromtheconvergencepointofview,theALmethodhasbasicallynodierencewithpenaltydecomposition.Butitismuchmoresuitableforourdeconvolutionframework,inwhichcanbexed,resultinginthesameinversekernelinalliterations. InverseKernelsforFastSpatialDeconvolution7 (a) (b)(c)(d)(e) Fig.2.Separatinglters.Aspatialinverseltershownin(a)canbeapproximatedasalinearcombinationofafewsimpleronesasshownfrom(b)-(e).Eachofthemisseparable.Thenallyrestoredimagein(a)canbeformedasalinearcombinationofimagesrestoredbythesesimpleltersrespectively.Byre-organizingtheterms,wegetanexpressionforthetargetimage:x=F 1 1 PjF(cj)2+ F(k)2!F 1 Xi F(cj)(F(zj) 1 F(\rj))+ F(k)F(y)!=w0@Xj2f1;2c0j(zj 1 \rj)+ k0y1A(16)wherewdenotesthesamespatialinverselterdenedinEq.(11).Thedier-enceisthatinthisformnolongervariesduringiterations.c0j(zj 1 \rj)canbeecientlycomputedusingforward/backworddierence.k0yisaconstantandcanbecomputedonlyoncebeforetheiteration.wisaspatialinversekernelthatcanalsobepre-computedandstored.Itseemsnowwesuccessfullyproduceworkableinversekernelwithoutheavycomputationspenttore-estimatingitineachiteration.Buttherearestillt-woaforementionedsizetheseparabilityissuesthatmayin\ruencedeconvolutioneciency.Wefurtherproposeadecompositionproceduretoaddressthem.InverseKernelDecompositionKerneldecompositiontechniqueshavebeenwidelyexplored.Steerablelters[7]decomposekernelsintolinearcombination 8L.Xu,X.TaoandJ.Jiaofasetofbasislters.Anotherkerneldecompositionisbasedonthesingularvaluedecomposition(SVD)ofwbytreatingitasamatrix[18].Comparedtosteerablelter,itisanon-parametricdecompositionforarbitrarylters.Givenourspatialkernelw,wedecomposeitasw=USV0,whereUandV0areunitaryorthogonalmatrices,V0isthetransposeofV,andthematrixSisaband-diagonalmatrixwithnonnegativerealnumbersinthediagonal.WeusefluandflvtodenotethelthcolumnvectorsofUandV,whichinessenceare1Dlters.wisexpressedasw=Xlslfluflv0(17)Convolvingwwithanimageisnowequivalenttoconvolvingasetof1Dkernelsfluandflv.Itcanbeecientlyappliedinspatialdomainwherethenumberofltersiscontrolledbythenon-zeroelementsinthesingularvaluematrix,inlinewiththerankofthekernel.Ifakernelisspatiallysmooth,whichiscommonfornaturalimages,therankcanbeverysmall.Itisthusallowedtouseonlyafew1Dkernelstoperformdeconvolution.Notethatwecanevenlowertheapproximationprecisionbydroppingsmallnon-zerossingularvaluesforfurtheracceleration.OneexampleofthekernelanditsdecompositionisshowninFig.2.Thelteredimagesareshowntogetherwiththeirseparablelters.Inthisexamples,7separableltersareusedtoapproximatetheinverseregularizedGaussian,whichveriesthatmostinversekernelsarenotoriginallyseparable.3.3MoreDiscussionsOurspatialdeconvolutionisaniterativeprocess.Foreachdeconvolutionprocess,weonlyneedtouseSVDtoestimatewasseveral1Dkernelsonce.Ifthekernelwasdecomposedbefore,wisstoredinourlesforquicklookup.Inthisregard,commonkernels,suchasGaussians,canbepre-computedtosavecomputationduringdeconvolution.Thespatialsupportofthe1Dinversekernelsdependsontheamountofregularization,i.e.theweight.Fornoisyimages,issetsmall,correspondingtostrongregularization.Accordingly,thesizeofinversekernelsissmall.Inpractice,thesupportof1Dkernelsisestimatedbythresholdinginsignicantvaluesinthekernelandremovingboundaryzerovalues,whicharedeterminedautomaticallyonceisgiven.Thepseudo-codeforinversekerneldeconvolutionisprovidedinAlg.1.4ExperimentalValidationWeevaluatethesystemperformancewithregardtorunningtimeandresultquality.Ourmainobjectiveistohandlefocal,Gaussianorevensparsemotionblur.Inourimplementation,primaryparametersinEq.(12)aresetasfollows:2[5003000],dependingontheimagenoiselevel;isxedto10forall InverseKernelsforFastSpatialDeconvolution9 Algorithm1x=FastSpatialDeconvolution(yk) 1w realF 11 PijF(ci)j2+ jF(k)j22fsl;flu;flvg svd(w)3Discardfsl;flu;flvgpairswithslbelowathreshold4x1 y,\r1 05fort=1tomaxIters6dozt+1i argminzE(xt;zi;\rti)7a i2f1;2gc0i(zt+1i 1 \rti)+ k0y8xt+1 09forl=1tolength(fslg)10doxt+1 xt+1+slafluflv011\rt+1i \rti (zt+1i cixt+1) Table1.Runningtime(inseconds)andPSNRsfordierentmethods ImageSize RL IRLS TVL1 FastPD Ours 325x365 0.91 85.83 7.50 0.59 0.57 1064x694 2.28 241.34 22.20 2.00 3.27 1251x1251 6.19 537.89 54.30 4.61 7.30 PSNRs 20.2 24.3 22.7 23.3 23.7 images;totally5iterationsareenoughinpractice.Wecomparedourmethodwithothers,includingthespatial-domainRichardson-Lucy(RL)deconvolution,IRLS[13](shortfortheiterativere-weightedleastsquares)approach,TVL1deconvolution[28]andthefastdeconvolution[11],denotedasPDfor\penaltydecomposition".TheTVL1methodisimplementedinClanguageandalltheotherfourmethodsareimplementedinMATLAB.Werun20iterationsforthestandardRL.Allothermethodsarebasedontheauthors'implementationwithdefaultparameters.Runningtimeisobtainedondierentsizesofimages.Intotal,wecollect10naturalimageswithdierentresolutions.TheyareblurredwithGaussianlterswithvariance2f12345grespectively.SmallGaussiannoiseisaddedtoeachimage.RunningtimeforthreeresolutionsisreportedinTable1.OurmethodissimilarlyfastasPDemployingFFTsandisamagnitudefasterthanIRLSandTVL1.Ourmethodupdateszwithanalyticalsolutions.Itcanbefurtherspedupbyusingalook-uptable.Aswispre-computed,wedonotincludeitsestimationtimeinthetable.Inourexperiments,a5151kerneliscomputedin0.1second.ThenalPSNRsofallthe10examplesareincludedinTable1.WeshowinFig.3avisualcomparisonalongwithclose-upsfordierentmethods.Ourresultiscomparablewiththesharpestonewhilenotcontainingextravisualartifacts. 10L.Xu,X.TaoandJ.Jia (a)Input(b)Richardson-Lucy(c)IRLS (d)TVL1(e)FastPD(f)OursFig.3.Visualcomparison.Similarqualityresultsmanifestthatourmethoddoesnotintroduceadditionalvisualartifacts. Fig.4.Samplemotionandfocalblurkernelsforvalidation.StatisticsofFiltersWenowpresentthestatisticsofthe1Dlterswlearnedfromdierenttypesofkernels.Wecollectedasetofltersinrealmotionblur,representativeGaussianconvolution,andnaturalout-of-focus.The8motionblurkernelsarefrom[14].TheGaussianblurkernelsarewithdierentscales,controlledbyvariance2f12345g.Wealsocollectfrominternettherealfocalblurkernels.Wenormalizeallofthemtosize3535.AfewexamplesareshowninFig.4.ThestatisticsinTable2indicatethatmotiondeconvolutiontypicallyrequiresmore1Dkernelstoapproximatetheinverselterthanothers,dueprimarilytolargekernelvariationandcomplexshapes.Convolvingtensofkernelsthatap-proximatewisinfactacompletelyparallelprocessandcanbeeasilyacceleratedusingmultiple-coreCPUandGPU.Thenumberof1Dkernelsisdeterminedbythresholdingthesingularvaluesanddroppingoutinsignicantones.Varyingthethresholdresultsindierentnumbersof1Dkernelsandthusaectstheperformance.WeshowinFig.5howthethresholdaectsthequalityofrestoredimages.Onethresholdcanbeappliedtodierenttypesofkernelstogeneratereasonableresults.Wealsonotebased InverseKernelsforFastSpatialDeconvolution11Table2.Kerneldecompositionstatistics.\Averagenumber"referstotheaveragenumberofnon-zerosingularvalues,i.e.,thenumberof1Dltersused.\Averagelength"isthelengthofeach1Dkernel. Type Avg.number Avg.length Motion 36.4 110.3 Gaussian 8.3 71.2 Out-of-focus 15.7 87.8 19212325272931 Gaussian Out-of-focus Motion PSNRSingularValueThreshold (log) Fig.5.PSNRsversussingularvaluethresholdsfordierenttypesofkernels.Thesin-gularvaluethresholdsareplottedinalogarithmicscale.onTable2thatonethresholdmaygeneratedierentnumbersof1Dkernelsdependingonthestructureandcomplexityoftheoriginalconvolutionkernels.5ApplicationsWeapplyourmethodtoafewcomputervisionandcomputationalphotographyapplications.5.1Deconvolution-IntensiveSuper-ResolutionIterativeback-projection[8]isoneeectiveschemetoupscaleimagesandvideos,andisfastingeneral.Inthisprocess,reconstructionerrorsarebackprojectedin-tothehighresolutionimagethroughinterpolationanddeconvolution,expressedasht+1=ht+(l (htG)#)"p;(18)whereGisakernelthatcouldbeGaussian[8]ornon-Gaussian[17],histhetargethigh-resolutionimageandlisitslow-resolutionversion.#and"aresimpledownscalingandupscalingwithinterpolationoperations.pisthepseudo-inverseofthekernel.Agoodppositivelyin\ruenceshigh-qualityimagesuper-resolution.Sowesubstituteourspatialdeconvolutionforp,whichcountsinregularization 12L.Xu,X.TaoandJ.Jia (a) Input(b) Back projection [8](c) Ours Fig.6.Super-resolutionbyback-projection.indeconvolution.ItproducestheresultsshowninFig.6.Theydemonstratetheusefulnessofourinversekernelscheme,asvisualartifactsaresuppressed.5.2ExtendedDepthofFieldTheproposedmethodcanbeappliedtoremovalofpartoffocalblur.Weemployitintheextendeddepthofeldphotography[12],whichgeneratesablurryimageforeachdepthlayerandrestoresitusingdeconvolution.Blurryimagegenerationisachievedbycontrollingthemotionofthedetectorduringimageintegrationorrotatingthefocusring.SincetheresultingblurPSFsbelongtothegeneralizedGaussianfamily,theycanbeecientlycomputedusingourspatialscheme.Fig.7showstwoexamples.Ittakes1.7sbyourmethodonasingleCPUcoretoproducetheresultsshownin(b)withresolution6811032.Incomparison,thefastdeconvolutionmethod[11]takes2stoproducetheresultsin(c).Ourmethodcanbefullyparallelizedtomuchspeedupcomputation.5.3MotionDeblurringMotionblurkernelsareingeneralasymmetric,correspondingtoalargernumberof1Dkernelsinourdecompositionstep.Itrevealsthenon-separablenatureofmotionkernels.Ourinversekernelschemeisstillapplicableherethankstotheindependenceofeach1Dlteringpass.WeshowinFig.8theIRLSdeconvolu-tionresultsof[13]andourinverselterresults.Thegroundtruthclearimagesandmotionblurkernelsarepresentedintheoriginalpaper[14].Whilebothap-proachesworkinspatialdomain,ourstakes0.5stoprocessthe255255images,comparedtothe70secondsbytheIRLSmethod.5.4Real-timePartialBlurRemovalOurmethoddirectlyhelpspartialimagedeconvolution.Fouriertransformre-quiressquareinputsandanyerrorproducedafterdomainswitchwillbeprop- InverseKernelsforFastSpatialDeconvolution13 (a) Input(b) Ours(c) PD [11] Fig.7.Reconstructedpicturesfromextendeddepthofeldcameras.agatedacrosspixelsduetothelackofspatialconsideration.Ourmethoddoesnothavetheseconstraints.Ourcurrentimplementationcanachievereal-timeperformanceon130130patchesonasingleCPUcore.Itisnotablethatanyshapesofregionscanbehandledinthissystem.Ourempiricallyprocessedre-gionsareslightlyexpandedfromtheusermarkedonestoincludemorepixelsinoptimizationinordertoavoidboundaryvisualartifacts.OneexampleisshowninFig.9,whereabookisfocalblurred.Werestoreapatchusingourmethod,whichdoesnotintroduceunexpectedringingartifacts.Ourmethodtakesonly0.07secondtoprocessthecontent,comparedwith0.4secondneededintheFFT-basedmethod[11]toprocessallpixelswithinthetightestboundingboxenclosingtheselectedregion.Theclose-upsareshownin(c)and(d).ThedierenceiscausedbyprocessingonlythemarkedpixelsbyourmethodandprocessingallpixelsintherectangularboundingboxbytheFFT-involvedmethod.6ConclusionWehavepresentedaspatialdeconvolutionmethodleveragingthepseudo-inversespatialkernelsunderregularization.FixedkernelestimationisachievedusingtheaugmentedLagrangianmethod.Ourframeworkisgeneralandndsmanyapplications.Itsimpactisthenumericalbridgetoconnectfastfrequency-domain 14L.Xu,X.TaoandJ.Jia (a) Input(b) IRLS [13](c) Ours Fig.8.Motiondeblurringexamples. (a)Input(b)Ourresult(c)Close-up(Ours)(d)Close-up(PD)Fig.9.PartialBlurRemoval.In(a),wemarkafewpixelsfordeconvolution.Theresultisshownin(b)withtheclose-upin(c).TheFFT-basedmethod(PD)yieldstheresultshownin(d)bydevolvingallpixelsintheboundingbox.operationsandrobustlocalspatialdeconvolution.Ourmethodinheritsthespeedandlocation-sensitivityadvantagesinthesetwostreamsofworkandopensupanewareaforfutureexploration.Themethodcouldbeamazinglyecientifthese1Dkernelbasesinvolvedindecompositionarehandledbydierentthreadsintheparallelcomputingarchitecture.ItworkswellforgeneralGaussianandotherpracticalmotionandfocalblurkernels.Onedirectionforfutureworkistoinvestigatespatiallyvaryinginversekernelsforcomplexblur.AcknowledgementsTheworkdescribedinthispaperwaspartiallysupportedbyagrantfromtheRe-searchGrantsCounciloftheHongKongSpecialAdministrativeRegion(ProjectNo.413113).TheauthorswouldliketothankShichengZhengforhishelpinimplementingpartofthealgorithm. InverseKernelsforFastSpatialDeconvolution15References1.Afonso,M.V.,Bioucas-Dias,J.M.,Figueiredo,M.A.:Anaugmentedlagrangianapproachtotheconstrainedoptimizationformulationofimaginginverseproblems.ImageProcessing,IEEETransactionson20(3),681{695(2011)2.Agrawal,A.K.,Raskar,R.:Resolvingobjectsathigherresolutionfromasinglemotion-blurredimage.In:CVPR(2007)3.Chakrabarti,A.,Zickler,T.,Freeman,W.T.:Analyzingspatially-varyingblur.In:CVPR.pp.2512{2519(2010)4.Cho,S.,Lee,S.:Fastmotiondeblurring.ACMTrans.Graph.28(5)(2009)5.Danielyan,A.,Katkovnik,V.,Egiazarian,K.:Imagedeblurringbyaugmentedla-grangianwithbm3dframeprior.In:WorkshoponInformationTheoreticMethodsinScienceandEngineering.pp.16{18(2010)6.Fergus,R.,Singh,B.,Hertzmann,A.,Roweis,S.T.,Freeman,W.T.:Removingcamerashakefromasinglephotograph.ACMTrans.Graph.25(3),787{794(2006)7.Freeman,W.T.,Adelson,E.H.:Thedesignanduseofsteerablelters.IEEETrans.PatternAnal.Mach.Intell.13(9),891{906(1991)8.Irani,M.,Peleg,S.:Motionanalysisforimageenhancement:Resolution,occlusion,andtransparency.JournalofVisualCommunicationandImageRepresentation4(4)(1993)9.Jerosch-Herold,M.,Wilke,N.,Stillman,A.,Wilson,R.:Magneticresonancequan-ticationofthemyocardialperfusionreservewithafermifunctionmodelforcon-straineddeconvolution.Medicalphysics25,73(1998)10.Joshi,N.,Zitnick,C.L.,Szeliski,R.,Kriegman,D.J.:Imagedeblurringanddenois-ingusingcolorpriors.In:CVPR.pp.1550{1557(2009)11.Krishnan,D.,Fergus,R.:Fastimagedeconvolutionusinghyper-laplacianpriors.In:NIPS(2009)12.Kuthirummal,S.,Nagahara,H.,Zhou,C.,Nayar,S.K.:Flexibledepthofeldphotography.IEEETrans.PatternAnal.Mach.Intell.33(1),58{71(2011)13.Levin,A.,Fergus,R.,Durand,F.,Freeman,W.T.:Imageanddepthfromacon-ventionalcamerawithacodedaperture.ACMTrans.Graph.26(3),70(2007)14.Levin,A.,Weiss,Y.,Durand,F.,Freeman,W.T.:Understandingandevaluatingblinddeconvolutionalgorithms.In:CVPR.pp.1964{1971(2009)15.Lin,Z.,Chen,M.,Ma,Y.:Theaugmentedlagrangemultipliermethodforexactrecoveryofcorruptedlow-rankmatrices.UIUCTechnicalReportUILU-ENG-09-2215(2010)16.Mathews,J.,Walker,R.L.:Mathematicalmethodsofphysics,vol.271.WABen-jaminNewYork(1970)17.Michaeli,T.,Irani,M.:Nonparametricblindsuper-resolution.In:ICCV(2013)18.Perona,P.:Deformablekernelsforearlyvision.IEEETrans.PatternAnal.Mach.Intell.17(5),488{499(1995)19.Raskar,R.,Agrawal,A.K.,Tumblin,J.:Codedexposurephotography:motiondeblurringusing\rutteredshutter.ACMTrans.Graph.25(3),795{804(2006)20.Richardson,W.:Bayesian-basediterativemethodofimagerestoration.JournaloftheOpticalSocietyofAmerica62(1),55{59(1972)21.Schmidt,U.,Rother,C.,Nowozin,S.,Jancsary,J.,Roth,S.:Discriminativenon-blinddeblurring.In:CVPR.pp.604{611.IEEE(2013)22.Shan,Q.,Jia,J.,Agarwala,A.:High-qualitymotiondeblurringfromasingleimage.ACMTrans.Graph.27(3)(2008) 16L.Xu,X.TaoandJ.Jia23.Shan,Q.,Li,Z.,Jia,J.,Tang,C.K.:Fastimage/videoupsampling.ACMTrans.Graph.27(5),153(2008)24.Starck,J.,Pantin,E.,Murtagh,F.:Deconvolutioninastronomy:Areview.Publi-cationsoftheAstronomicalSocietyofthePacic114(800),1051{1069(2002)25.Tai,Y.W.,Lin,S.:Motion-awarenoiselteringfordeblurringofnoisyandblurryimages.In:CVPR.pp.17{24(2012)26.Tikhonov,A.,Arsenin,V.,John,F.:Solutionsofill-posedproblems(1977)27.Wiener,N.:Extrapolation,interpolation,andsmoothingofstationarytimese-ries:withengineeringapplications.JournaloftheAmericanStatisticalAssociation47(258)(1949)28.Xu,L.,Jia,J.:Two-phasekernelestimationforrobustmotiondeblurring.In:ECCV(1).pp.157{170(2010)29.Xu,L.,Zheng,S.,Jia,J.:Unnaturall0sparserepresentationfornaturalimagedeblurring.In:CVPR.pp.1107{1114(2013)30.Yuan,L.,Sun,J.,Quan,L.,Shum,H.Y.:Imagedeblurringwithblurred/noisyimagepairs.ACMTrans.Graph.26(3),1(2007)31.Yuan,L.,Sun,J.,Quan,L.,Shum,H.Y.:Progressiveinter-scaleandintra-scalenon-blindimagedeconvolution.ACMTrans.Graph.27(3)(2008)32.Zhou,C.,Lin,S.,Nayar,S.K.:Codedaperturepairsfordepthfromdefocusanddefocusdeblurring.InternationalJournalofComputerVision93(1),53{72(2011)