简介:Anadaptiveinterferencecancleralgorithmbasedonfourth-ordercumulantisproposedtoovercomethefaultofconventionalautocorrelation-basedLeastMeanSquare(LMS)algorithmwhichcannotsuppressthecorrelatedGaussiannoises.Therecursiveformulationoftheproposedalgorithmisgiventoreduceitscomputationalcomplexity.Thesimulationresultsarepresentedtodemonstratetheeffectivenessofthisnovelapproach.
简介:Operationalsoftwaresystemsoftenexperiencean'aging'phenomenon,characterizedbyprogressiveperformancedegradationandasuddenhang/crashfailure.Softwarerejuvenationisaproactivefault-tolerancestrategyaimedtopreventunexpectedoutagesduetoaging.Anewrejuvenationstrategybasedonmeasurementandtime,andconstructsaSRNmodeltoevaluatethisstrategy.Numericalresultsshowthatthisstrategyoutweighseitherthepurelymeasurementbasedorthepurelytime-basedstrategyandcaneffectivelyimprovesystemperformance.
简介:Recentprogressesmadebyauthorsonmonochromaticandtunableterahertz(THz)generationbasedonnonlinearopticsarereviewed,includingTHzparametricoscillation(TPO)anddifferencefrequencygeneration(DFG).Fromthetechnicalpointofview,wedevelopextra-andintra-cavitysurface-emittedTPO,aswellasDFGwithQPM-GaAscrystal.Fromthepointofviewofmechanism,Cherenkovphase-matchingiscomprehensivelyinvestigatedinbothbulkcrystalandplanarwaveguide.AnovelschemeforcascadingenhancedCherenkovDFGinwaveguideisproposed.Fromthepointofviewofmaterial,organiccrystal4-N,N-dimethylamino-4’-N’-methyl-stibazoliumtosylate(DAST)isutilizedasthenonlinearmedium.
简介:Statisticalshapepriormodelisemployedtoconstructthedynamicsinprobabilisticcontourestimation.Byapplyingprincipalcomponentanalysis,plausibleshapesamplesareefficientlygeneratedtopredictcontoursamples.Basedontheshape-dependentdynamicsandprobabilisticimagemodel,aparticlefilterisusedtoestimatethecontourwithaspecificshape.Comparedwiththedeterministicapproachwithshapeinformation,theproposedmethodissimpleyetmoreeffectiveinextractingcontoursfromimageswithshapevariationsandocclusion.
简介:FractionalBrownianmotion,continuouseverywhereanddifferentiablenowhere,offersaconvenientmodelingforirregularnonstationarystochasticprocesseswithlong-termdependenciesandpowerlawbehaviorofspectrumoverwiderangesoffrequencies.Itshowshighcorrelationatcoasescaleandvariesslightlyatfinescale,whichissuitableforandsuccessfulindescribingandmodelingnaturalscenes.Ontheotherhand,man-madeobjectscanbeconstructivelywelldescribedbyusingasetofregularsimpleshapeprimitivessuchasline,cylinder,etc.andarefreeoffractal.Basedonthedifference,weprovideamethodtodiscriminateman-madeobjectsfromnaturalscenes.Experimentsareusedtodemonstratedthegoodefficiencyofthedevelopedtechnique.
简介:TherearetwomajorapproachesforBlindSignalSeparation(BSS)problem:MaximumEntropy(ME)andMinimumMutualInformation(MMI)algorithms.BasedontherecursivearchitectureandtherelationshipbetweentheMEandMMIalgorithms,anExtendedME(EME)algorithmisproposedbyusingprobabilitydensityfunction(pdf)estimationoftheoutputstodeducethecorrespondingiterativeformulasinBSS.Basedonthesimulationresults,itcanbeconcludedthattheproposedalgorithmhasbetterperformancesthanthetraditionalMEalgorithminconvolutemixtureBSSproblems.
简介:Accordingtothecharacteristicsofneuralsignal,alow-voltageandhigh-speedoperationaltransconductanceamplifierhasbeenrealizedasadirectreadoutcircuit.Toreducethesizeandcostofthischip,theamplifierusesfeedforwardcompensationtechniquewithoutMillercapacitor.Usingcapacitorsasaclosed-loopfeedbacknotonlyamplifiestheneuralsignal,butalsoreduceschippowerconsumptionbyreplacingthefeedbackresistor.Simulationswerecarriedoutwith0.18mCMOStechnology,andthesimulationresultssatisfiedtherequirementsofneuralsignal.
简介:Thepaperdescribesatexture-basedfasttextlocationschemewhichoperatesdirectlyintheDiscreteWaveletTransform(DWT)domain.Bythedistinguishingtexturecharacteristicsencodedinwavelettransformdomain,thetextisfastdetectedfromcomplexbackgroundimagesstoredinthecompressedformatsuchasJPEG2000withoutfulldecompress.Comparedwithsometraditionalcharacterlocationmethods,theproposedschemehastheadvantagesoflowcomputationalcost,robusttosizeandfontofcharactersandhighaccuracy.Preliminaryexperimentalresultsshowthattheproposedschemeisefficientandeffective.
简介:Basedontheprinciplesofthevisualcryptographyandthelawoflargenumbers,theunexpandedsharesaregeneratedduringtheprocessesofembeddingandverifyingthehiddenwatermark.Thewatermarkembeddingisdoneinthefrequencydomain,whichcanbedecodedbythehumanvisualsystem(HVS)withoutthenecessityofanycomplicatedcomputationandthehelpoftheoriginalimage.Experimentalresultsindicatedthatourmethodhadagoodrobustnessondarkening,lightening,blurring,sharpening,noise,distorting,jitter,jointphotographicexpertsgroup(JPEG)compression,andcropattacks.
简介:Theperformanceofadaptivebeamformingtechniquesislimitedbythenonhomogeneouscluttersce-nario.AnaugmentedKrylovsubspacemethodisproposed,whichutilizesonlyasinglesnapshotofthedataforadaptiveprocessing.ThenovelalgorithmputstogetheradatapreprocessorandadaptiveKrylovsubspacealgo-rithm,wherethedatapreprocessorsuppressesdiscreteinterferenceandtheadaptiveKrylovsubspacealgorithmsuppresseshomogeneousclutter.Thenovelmethodusesasinglesnapshotofthedatareceivedbythearrayantennatogenerateacancellationmatrixthatdoesnotcontainthesignalofinterest(SOI)component,thus,itmitigatestheproblemofhighlynonstationaryclutterenvironmentandithelpstooperateinreal-time.Thebenefitofnotrequiringthetrainingdatacomesatthecostofareduceddegreeoffreedom(DOF)ofthesystem.Simulationillus-tratestheeffectivenessincluttersuppressionandadaptivebeamforming.Thenumericresultsshowgoodagreementwiththeproposedtheorem.
简介:我们构造稀少的表示和subspace表示的一个合作模型。首先,我们代表追踪在原则部件分析(PCA)指向subspace,然后我们采用L1规则化限制剩余学期的稀少,限制表示系数的稀少的一个L2规则化学期,并且限制在重建和目标之间的距离的一个L2标准。然后,我们在粒子过滤器框架实现算法。而且,一个反复的方法被介绍得到剩余和系数的全球最小。最后,一个其他的模板更改计划被采用避免被不精密的更改引起的追踪的飘移。在实验,我们在9个序列上测试算法,并且把结果与5个state-of-art方法作比较。根据结果,我们能断定我们的算法比另外的方法更柔韧。
简介:MotionEstimationBasedonGlobalandLocaCompensabilityAnalysisTXMotionEstimationBasedonGlobalandLocalCompensabilityAnalysisFanJin...
简介:FractionalBrownianmotion,continuouseverywhereanddifferentiablenowhere,offersaconvenientmodelingforirregularnonstationarystochasticprocesseswithlong-termdependenciesandpowerlawbehaviorofspectrumoverwiderangesoffrequencies.Itshowshighcorrelationatcoarsescaleandvariesslightlyatfinescale,whichissuitableforandsuccessfulindescribingandmodelingnaturalscenes.Ontheotherhand,man-madeobjectscanbeconstructivelywelldescribedbyusingasetofregularsimpleshapeprimitivessuchasline,cylinder,etc.andarefreeoffractal.Basedonthedifference,amethodtodiscriminateman-madeobjectsfromnaturalscenesisprovided.Experimentsareusedtodemonstratethegoodefficiencyofdevelopedtechnique.
简介:Aimingatthetraditionalpassivedeceptionmodels,thispaperconstructsaDecoyPlatformbasedonIntelligentAgent(DPIA)torealizedynamicdefense.Thepaperexploresanewdynamicdefensemodelbasedonactivedeception,introducesitsarchitecture,andexpatiatesoncommunicationmethodsandsecurityguaranteeininformationtransference.SimulationresultsshowthattheDPIAcanattracthackeragilityandactivity,leadabnormaltrafficintoit,distributealargenumberofattackdata,andensurerealnetworksecurity.
简介:
简介:Inthispaper,weproposeandexperimentallydemonstrateamethodtorealizeapolarizationdetectorbyusingacommerciallyavailablepolymethylmethacrylate(PMMA)pipe.Anasymmetricpolarization-relatedstructureisfabricatedbyetchingdifferentgratingstructuresontheexternalsurfaceofthepipe.Theresultsshowthatwhenthepiperotatesupto90°,thepeakofthetransmissionspectrumcanevidentlyshifttowardsthelowfrequency,whichmeansthepipewithgratingstructurescanserveasapolarizationdetector.Thisdesignpossessesthemeritsofasimplestructureandeasyfabrication.
简介:Thequantumtheoryapplicationisahotresearchareainrecentyears,especiallythetheoryofquantummechanics.Inthispaper,wefocusontheresearchofimagesegmentationbasedonquantummechanics.Firstly,thetheoryofquantummechanicsisintroduced;afterwards,areviewofimagesegmentationmethodsbasedonquantummechanicsispresented;andfinally,thecharacteristicsaboutthequantummechanicsappliedtoimageprocessingareconcluded.Twomainresearchtopicsarediscussedinthispaper.Oneistoemphasizethatquantummechanicscanbeappliedindifferentresearchareas,suchasimagesegmentation,andthesecondistoconcludesomemethodsinimagesegmentationandgivesomesuggestionsforpossiblenovelmethodsbyapplyingquantummechanicstheory.Asasummary,thisisareviewpaperwhichpresentssomemethodsbasedonthefeasibletheoryinquantummechanicsaimingatachievingabetterperformanceinimagesegmentation.