简介：Asimpleandadaptivelosslesscompressionalgorithmisproposedforremotesensingimagecompression,whichincludesintegerwavelettransformandtheRiceentropycoder.ByanalyzingtheprobabilitydistributionofintegerwavelettransformcoefficientsandthecharacteristicsofRiceentropycoder,thedivideandrulemethodisusedforhigh-frequencysub-bandsandlow-frequencyone.High-frequencysub-bandsarecodedbytheRiceentropycoder,andlow-frequencycoefficientsarepredictedbeforecoding.Theroleofpredictoristomapthelow-frequencycoefficientsintosymbolssuitablefortheentropycoding.ExperimentalresultsshowthattheaverageCompressionRatio(CR)ofourapproachisabouttwo,whichisclosetothatofJPEG2000.Thealgorithmissimpleandeasytobeimplementedinhardware.Moreover,ithasthemeritsofadaptability,andindependentdatapacket.Sothealgorithmcanadapttospacelosslesscompressionapplications.

简介：Oneofthemostcommonimageprocessingtasksinvolvestheremovalofnoisefromimages.Noisecanbeintroducedduringimagecapture,duringtransmission,orduringstorage.Fordesignpurposes,noisesourcesarefrequentlyapproximatedbyrandomvariableswithaknownprobabilitydistribution.Onecommonnoisemodelcorruptsasignalbyintroducingimpulses.Andthesurfaceoftheimagedisturbedbyimpulsenoisedisplaysmanypeaksorvales.Accordingtothecharacteristicofimpulsenoise,anovelalgorithmisproposedtothedetectionofimpulsenoisepointfromimagesbasedondirectionalderivatives.First,thetheoryofcalculusondirectionalderivativesisintroducedindetail.Thenitisappliedtothefieldofimagetoremovingnoisewiththediscreteformderivedfromitscontinuousmathematicalmodel.Andanumberofcontrastingsimulationsillustratethatouralgorithmnotonlycanpreservethestructureinformationwhileremovingimpulsenoisebutalsocanmostlysavethegrayvalueofthepixelsundisturbedbynoise.Inaddition,thecomparisonsofthefilteringperformanceforremovingimpulsenoiseareanalyzedindetailinthecaseofdifferentnoisedensities,andalsoshowthatthealgorithmsuggestedoutperformstheconventionalfilteralgorithmssuchasmeanfilter,medianfilterandsooninspeedandimpulsenoisereduction,especiallyinrandom-valuedimpulsenoisereduction.Soitisaverygoodalternativetotheexistingschemes.

简介：Inthispaper,thetechniqueofquasi-losslesscompressionbasedontheimagerestorationispresented.Thetechniqueofcompressiondescribedinthepaperincludesthreesteps,namelybitcompression,correlationremovingandimagerestorationbasedonthetheoryofmodulationtransferfunction(MTF).Thequasi-losslesscompressioncomestoahighspeed.Thequalityofthereconstructionimageunderrestorationisuptoparofthequasi-losslesswithhighercompressionratio.TheexperimentsoftheTMandSPOTimagesshowthatthetechniqueisreasonableandapplicable.

简介：Manyapproachesforhigh-resolutionimagereconstructionhavebeenproposedinsomeliteratures.Oneofthemostcommonlywaysistoreconstructahigh-resolutionimagefromanumberofrotatedandtranslatedimageswithlowresolution.Inthisprocess,theexposuredifferenceamongoriginalimageswilldecreasethequalityofthereconstructedimage.Inordertoremovetheinfluenceoftheexposuredifference,alightenergymatchingmethodisproposedinthispaper.Thetheoreticalanalysisisillustratedindetails.Experimentalresultsshowthatthetheoreticalanalysisiscorrectandtheproposedmethodisvalid.

简介：

简介：

简介：

简介：多尺度的策略在基于波形的速度倒置的成功的申请是很重要的。策略顺序preceeds从长当模特儿突然速度可伸缩，很好在完整的波形倒置(FWI)被开发了解决本地mininum问题。相反，它的不好在图象域波形断层摄影术(IWT)理解，哪个幕後放映的影像普通图象的支离破碎的波形部件聚在一起进速度更改。IWT对本地最小的问题不太敏感，但是趋于与低分辨率造长规模的模型。以便由IWT造长规模、短规模的模型，我们讨论在图象域限制的几多尺度的策略。策略包括模型reparameterization，客观功能切换和重新可伸缩的坡度。Marmsousi模型和真实数据上的数字测试证明我们的建议多尺度的IWT在有宽wavenumber光谱的buidling速度模型是有效的。

简介：有人的脸的图象在成像领域包括必要部分。在面部部分的吸藏或损坏将带显著不快和信息损失。我们建议能修理的一个算法堵塞了或自动地损坏了面部图象，命名‘facial图象inpainting'。Inpainting是处理方法恢复失踪的图象部分的一套图象。我们在由介绍面部领域知识油漆方法扩大图象。与一个脸数据库的支持，我们的途径宣传结构的信息，即，特征点和边地图，从类似的脸到失踪的面部区域。把推断的结构的信息用作指导，在油漆算法的一幅基于模范的图象被采用从来源部分在一样的脸拷贝补丁到失踪的部分。在油漆的面部图象的这个最新建议的概念超过在美容从一个脸数据库塑造，并且由从一样的脸图象推断颜色和质地从不同图象处于成像条件避免变化的问题的由宣传的画方法传统。我们的系统生产是几乎没看见的缺点的无缝的脸。

简介：

简介：Wavelet-fractalbasedSAR(syntheticapertureradar)imageprocessingisoneoftheadvancedtechnologiesinimageprocessing.Themainconceptofanalysisisthatafterwavelettransformation,multifractalspectrumofthesignalisdifferentfromthatofnoise.ThisdifferenceisusedtoalleviatethenoiseproducedbySARimage.ThemethodtodenoiseSARimageusingtheprocessbasedonwavelet-fractalanalysisisdiscussedindetail.Essentially,thepresentmethodfocusesonadjustingtheHlderexponentαofmultifractalspectrum.Aftersimulation,αshouldbeadjustedto1.72-1.73.Themorethevalueofαexceeds1.73,thelessdistinctivetheedgesofSARimagebecome.Accordingtotheauthorsdenoisingisoptimalatα=1.72-1.73.Inotherwords,whenα=1.72-1.73,asmoothanddenoisedSARimageisproduced.

简介：一个信号处理方法为高速度在水下图象的声学的传播被介绍。它有二部分。第1部分介绍信号处理为在水下声学的协调通讯。第1部分包括3个技术的点。(1)Doppler移动compensation.Chirp信号在数据包裹之间被插入。在二个拷贝关联函数之间的关联进程给吝啬的Doppler移动的更多的精确评价。然后，它能被采样补偿数据。在适应决定反馈均衡器(DFE)一个适应阶段赔偿者与快，自我优化的最不吝啬的广场(FOLMS)改编算法被利用与第二顺序阶段锁环算法比赔偿者导致更好的运动忍耐。吝啬的Doppler移动赔偿和适应阶段赔偿者的联合的表演相当好。(2)差异combiner(DC)在均衡器之前使用了。combiner和适应DFE基于FOLMS改编算法。这导致减少的计算复杂性和更好的性能。(3)串联了均衡器，涡轮架子编码了调整(TCM)解码器和重复算法。一个新bitsymbol变换器基于软产量Viterbi算法(SOVA)被学习。与传统的决定作比较，编码并且印射算法，在将近2的新变换器罐头还原剂小点错误率(BER)订。第2部分主要在一个柔韧的图象压缩算法附近。基于分离小浪变换并且修理了长度编码，为声学的图象的一个柔韧的压缩算法被学习。算法包括4个技术的点。(1)利用小浪基于转变图象的CDF9/7。(2)分析subband系数的精力分发。合适的转变层数字是3。(3)根据他们的精力分发把不同量子化步用于不同subbands。(4)使用修理了长度编码阻止错误繁殖。结果证明算法在图象质量，压缩率，和大多数之中完成平衡重要，坚韧性到BER。灰色的规模的压缩的位率声学的图象是0.85小点/象素。当BER比10-3低时，图象质量仍然保持好。当BER起床到10-2时，在那里削减一些小脏点。基于提及的上面的信号处理技术，一在水下声学的通讯系统是运作的频率乐队是的

简介：Thisarticlepresentsanovelimageinterpolationbasedonrationalfractalfunction.Therationalfunctionhasasimpleandexplicitexpression.Atthesametime,thefractalinterpolationsurfacecanbedefinedbyproperparameters.Inthispaper,weusedthemethodof‘coveringblanket’combinedwithmulti-scaleanalysis;thethresholdisselectedbasedonthemulti-scaleanalysis.Selectingdifferentparametersintherationalfunctionmodel,thetextureregionsandsmoothregionsareinterpolatedbyrationalfractalinterpolationandrationalinterpolationrespectively.Experimentalresultsonbenchmarktestimagesdemonstratethattheproposedmethodachievesverycompetitiveperformancecomparedwiththestate-of-the-artinterpolationalgorithms,especiallyinimagedetailsandtexturefeatures.

简介：AmixedschemebasedonWaveletTransformation(WT)isproposedforimageedgedetection.TheschemecombinesthewavelettransformandtraditionalSobelandLoG(LaplacianofGaussian)operatoredge-detectionalgorithms.Theprecisetheoryanalysisisgiventoshowthatthewavelettransformationhasanadvantageforsignalprocessing.SimulationresultsshowthatthenewschemeisbetterthanonlyusingtheSobelorLoGmethods.Complexityanalysisisalsogivenandtheconclusionisacceptable,thereforetheproposedschemeiseffectiveforedgedetection.

简介：Onthebasisofstereoimageanalysis,thechangedetectionofman-madeobjectsinurbanareasisintroduced.Informationoftheheightofman-madeobjectscanbeappliedtoreinforcetheirchangedetection.BycomparisonbetweenthenewandoldDSMs,thechangedregionsareextracted.However,ouraimistodetectchangesofman-madeobjectsinurbanareaandfurtherinthepotentialareasbythemeansofline-featurematchingandgradientdirectionhistogram.TheexperimentsbasedontheaerialimagesfromJapanhaveproventhatthealgorithmiscorrectandefficient.

简介：Anovelimagestabilizationsystemispresented,whichconsistsofaglobalfeaturepointtrackingbasedmotionestimation,aKalmanfilteringbasedmotionsmoothingandanimagemosaicbasedpanoramiccompensation.Theglobalmotionisestimatedusingfeaturepointmatchinganditerationwiththeleast-squaremethod.Then,theKalmanfilterisappliedtosmooththeoriginalmotionvectorstoeffectivelyalleviateunwantedcameravibrationsandfollowtheintentionalcamerascan.Lastly,thelossinformationofim...

简介：Flowcytometryandimageanalysistechniquewereusedtoquantltatethenucleiofvarioussofttissuetumors.Asinglerepresentingsectionfromsofttissuesarcomawasusedforhistologicgrading.Histologlcandcytometriccomparativeanalysesshowedthatall21benigntumorswerediploid.Among62casesofsofttissuesarcoma,45(73%)wereaneuploid.Therewasasignificantdifferenceinthenuclearareabetweenbenignandmalignanttumors(P<0.01),dlploidandaneuploidtumors(P<0.05).ThetwonewtechniquesarevaluableIncellularquantitativemeasurementforsofttissuetumors.

简介：MostofImageQualityAssessment(IQA)metricsconsistoftwoprocesses.Inthefirstprocess,qualitymapofimageismeasuredlocally.Inthesecondprocess,thelastqualityscoreisconvertedfromthequalitymapbyusingthepoolingstrategy.Thefirstprocesshadbeenmadeeffectiveandsignificantprogresses,whilethesecondprocesswasalwaysdoneinsimpleways.Inthesecondprocessofthepoolingstrategy,theoptimalperceptualpoolingweightsshouldbedeterminedandcomputedaccordingtoHumanVisualSystem(HVS).Thus,areliablespatialpoolingmathematicalmodelbasedonHVSisanimportantissueworthyofstudy.Inthispaper,anewVisualPerceptualPoolingStrategy(VPPS)forIQAispresentedbasedoncontrastsensitivityandluminancesensitivityofHVS.ExperimentalresultswiththeLIVEdatabaseshowthatthevisualperceptualweights,obtainedbytheproposedpoolingstrategy,caneffectivelyandsignificantlyimprovetheperformancesoftheIQAmetricswithMeanStructuralSIMilarity(MSSIM)orPhaseQuantizationCode(PQC).ItisconfirmedthattheproposedVPPSdemonstratespromisingresultsforimprovingtheperformancesofexistingIQAmetrics.

简介：Studyofimageenhancementshowsthatthequalityofimageheavilyreliesonhumanvisualsystem.Inthispaper,weapplythisfacteffectivelytodesignanewimageenhancementmethodformedicalimagesthatimprovesthedetailregions.First,theeyeregionofinterest(ROI)issegmented;thentheUn-sharpMasking(USM)isusedtoenhancethedetailregions.Experimentsshowthattheproposedmethodcaneffectivelyimprovetheaccuracyofmedicalimageenhancementandhasasignificanteffect.

简介：Airborne3DimagewhichintegratesGPS,attitudemeasurementunit(AMU),scanninglaserrangefinder(SLR)andspectralscannerhasbeendevelopedsuccessfully.ThespectralscannerandSLRusethesameopticalsystemwhichensureslaserpointtomatchpixelseamlessly.Thedistinctiveadvantageof3Dimageisthatitcanproducegeo_referencedimagesandDSM(digitalsurfacemodels)imageswithoutanygroundcontrolpoints(GCPs).ItisnolongernecessarytosurveyGCPsandwithsomesoftwaresthedatacanbeprocessedandproducedigitalsurfacemodels(DSM)andgeo_referencedimagesinquasi_real_time,therefore,theefficiencyof3Dimageis10～100timeshigherthanthatoftraditionalapproaches.Theprocessingprocedureinvolvesdecomposingandcheckingtherawdata,processingGPSdata,calculatingthepositionsoflasersamplepoints,producinggeo_referencedimage,producingDSMandmosaicingstrips.　Theprincipleof3Dimageisfirstintroducedinthispaper,andthenwefocusonthefastprocessingtechniqueandalgorithm.Theflighttestsandprocessedresultsshowthattheprocessingtechniqueisfeasibleandcanmeettherequirementofquasi_real_timeapplications.