简介:Formanyclusteringalgorithms,itisveryimportanttodetermineanappropriatenumberofclusters,whichiscalledclustervalidityproblem.Inthispaper,anewclusteringvalidityassessmentindexisproposedbasedonanovelmethodtoselectthemarginpointbetweentwoclustersforin-ter-clustersimilaritymoreaccurately,andprovidesanimprovedscatterfunctionforintra-clustersimilarity.Simulationresultsshowtheeffectivenessoftheproposedindexonthedatasetsunderconsiderationregardlessofthechoiceofaclusteringalgorithm.
简介:Byusingfourspeciallydesignednarrowbandpassfiltersandphotodetectorsintheinstrument,theincidentandreflectedradiancesofsunlightonthevegetationareopticallysensed,attheredandnearinfraredbands,thenthenormalizeddifferencevegetationindex(NDVI)isprocessedbyamicroprocessor.ComparedwithconventionalspectrometermeasuringmethodofNDVI,theinstrumentiseasytobeused,compact,lightandlow-cost.
简介:Aimingattheproblemthatthereisn'tanystandardizedcalibrationmethodfortheperformanceindexofsurfaceplasmonresonance(SPR)sensor,thecalibrationmethodforkeyperformanceindicesofSPRsensorissummarizedandproposedbasedonthecomparisonofrelativemethodsanddefinitionofeachindex.Experimentaldataofsucrosesolutionswithvariousconcentrationsareobtainedbytheself-buildingSPRinstrument,andthenthecalibrationmethodisusedtodeterminetheperformanceindices,suchasnoise,drift,sensitivity,resolution,linearity,dynamicrangeandreproducibility.Experimentalresultsshowthatthedefinitionofindicesisreasonable,andthecalibrationmethodiscorrect,whichhasgreatsignificanceforperformanceevaluationofSPRsensor.
简介:Aswitchingvariabilityindex(SVI)constantfalsealarmrate(CFAR)detectorisproposedforimprovingthedetectionperformanceofVI-CFARdetectorsinmultipletargetsbackgrounds.Whenthepresenceofnon-homogeneityinCFARreferencewindowsisindicatedbyaVI-CFARdetector,aswitchingCFARdetectorisintroducedtooptimizetheperformanceoftheVI-CFARdetectorinhomogeneous,multipletargetsandclutteredgebackgrounds.ThestructureandparametersselectionmethodoftheSVI-CFARdetectorispresented.ComparisonswithclassicCFARdetectorsandrecentlyproposeddetectorsarealsogiven.TheoreticalanalysisandsimulationresultsshowthatSVICFARdetectormaintainsthegoodperformanceoftheVI-CFARdetectorinhomogeneousandclutteredgebackgrounds,whilegreatlyimprovingthecapacityofanti-multitargets.
简介:Losslessdatahidingcanrestoretheoriginalstatusofcovermediaafterembeddedsecretdataareextracted.In2010,Wangetal.proposedalosslessdatahidingschemewhichhidessecretdatainvectorquantization(VQ)indices,buttheencodingstrategiesadoptedbytheirschemeexpandthefinalcodestream.ThispaperdesignsfourembeddingandencodingstrategiestoimproveWangetal.'sscheme.TheexperimentresultoftheproposedschemecomparedwiththatoftheWangetal.'sschemereducesthebitratesofthefinalcodestreamby4.6%andraisesthepayloadby1.09%onaverage.
简介:ByuseoftheKeldyshnn-equilibriumGreen's-functionmethods,thethirdharmonicsusceptibilitiesoftwopolyanilinefamilies,PANI-HCl巨PANI-H3PO4,arccalculated[x^(3)(ω)≈10^-12esu].Itwasfoundthatthethirdharmonicsusceptibilityofpolyanilinestronglydependsonthedelocalizationoftheelectrons.Therefractiveindicesn(λ=589nm)ofPANI-HClandPANI-H3PO4arecalculatedbyuseofthreecommonmethods(theLorentz-Lorentztheoreticalmodel,theGladstone-DalegroupcontributionandtheVogelgroupcorrelation)basedongroupcontributionstomolarrefraction.Thecalculatenvaluesarevariedfrom1.31to1.42forPANI-HCland1.36to1.45forPANI-H3PO4.
简介:Thecharacteristicmodificationsarereportedonthesurfaceofpolymericwaveguidefilmintheprocessofvolume-gratingfabrication.Thelightfromamode-locked76MHzfemtosecondlaserwithpulsedurationof200fsandwavelengthof800nmisfocusednormaltothesurfaceofthesample.Thesurfacemorphologymodificationsareascribedtoafactthatsurfaceswellingoccursduringtheprocess.Periodicmicro-structureisinscribedwithincreasingincidentpower.Thelaser-inducedswellingthresholdonthegrating,whichishigherthanthatoftwo-photoninitiatedphoto-polymerization(TPIP)(8mW),isverifiedtobeabout20mW.Itisfeasibletoenhancethesurfacesmoothnessofintegratedopticsdevicesforfurtherencapsulation.Thevariationofmodulationdepthisstudiedfordifferentvaluesofincidentpowerandscanspacing.Ablationaccompaniedwithsurfaceswellingappearswhenthepowerishigher.Byoptimizingthelasercarvingparameters,highlyefficientgratingdevicescanbefabricated.
简介:Inthispaper,thephotonicbandgapofphotoniccrystalswithchiralandnegative-indexmaterialsalternatelyisinvestigated.Itisshownthatthiskindofphotoniccrystalexhibitsverywidephotonicbandgap.Thephotoniccrystalswithadefectlayerhaveanarrowdefectmodeinthebandgap,whichcanbeusedasaverygoodfilteringmaterialbychoosingtheparametersofthedefectlayerproperly.
简介:Inthispaper,aplasmonicsrefractiveindexsensorwiththesemiring-stubsystemcoupledtoabuswaveguideispro-posed.Thestructurecanachievewavelengthbandtunableandultra-sharpFanoresonance.Thecouplingmechanismbetweenthebrightmodeanddarkmodeisinvestigatedindetail,andapparentplasmoninducedtransparency(PIT)isrealizedontheresonancewavelength.Thefullwidthathalfmaximum(FWHM)oftheresonantwavelengthismodulatedbyadjustingthecoupleddistance.Thesensitivityandfigureofmerit(FOM)oftheproposedsensor'canreachupto600nm/RIUand120onthevisibleregion,respectively,andthisperformancecanbehelpfulfordesigningthephotonicintegratedcircuitandopticalcommunication,inaddition,itcanbeappliedforbio-sensing.
简介:Lightpropagationinnanoparticle-loadedencapsulantsissimulatedbasedonthemethodofMonteCarlo,referringtothemultilayergraded-refractive-indexstructureforLEDencapsulants.Andtheinfluenceofscatteringcoefficientonthetransmittanceisanalyzed.Theresultsshowthatthetransmittancedecreaseswithscatteringcoefficient.Foragivennumberoflayers,theencapsulantswillyieldmaximumtransmittanceiftherefractiveindexvalueofeachlayerisoptimized.Thenanoparticle-loadedencapsulantconsistsofmultiplelayerswithrefractive-indexvaluesthatgraduallydecrease.Thetransmittancewillbehigherthanthatoftraditionalnon-gradedencapsulants,andwillimprovelightextractionefficiencyofLED.
简介:Torealizecontent-basedretrievaloflargeimagedatabases,itisrequiredtodevelopanefficientindexandretrievalscheme.ThispaperproposesanindexalgorithmofclusteringcalledCMA,whichsupportsfastretrievaloflargeimagedatabases.CMAtakesadvantagesofk-meansandself-adaptivealgorithms.Itissimpleandworkswithoutanyuserinteractions.Therearetwomainstagesinthisalgorithm.Inthefirststage,itclassifiesimagesinadatabaseintoseveralclusters,andautomaticallygetsthenecessaryparametersforthenextstage-k-meansiteration.TheCMAalgorithmistestedonalargedatabaseofmorethantenthousandimagesandcompareitwithk-meansalgorithm.Experimentalresultsshowthatthisalgorithmiseffectiveinbothprecisionandretrievaltime.