ealsousedtoachievesimilarresults.Probabilisticmodelsusingthestatisticalpropertiesofthewaveletcoefficientseemedtooutperformthethresholdingtechniquesandgainedground.Recently,muchefforthasbeendevotedtoBayesiande-noisinginWaveletdomain.HiddenMarkovModelsandGaussianScaleMixtureshavealsobecomepopularandmoreresearchcontinuestobepublished.TreeStructuresorderingthewaveletcoefficientsbasedontheirmagnitude,scaleandspatiallocationhavebeenresearched.DataadaptivetransformssuchasIndependentComponentAnalysis(ICA)havebeenexploredforsparseshrinkage.Thetrendcontinuestofocusonusingdifferentstatisticalmodelstomodelthestatisticalpropertiesofthewaveletcoefficientsanditsneighbors.Futuretrendwillbetowardsfindingmoreaccurateprobabilisticmodelsforthedistributionofnon-orthogonalwaveletcoefficients..Classificat平,开始阈值)和小波分析选择对收缩率影响较大成功程序。
&