EEG-signal-processing-脑电信号处理方法算法VIP免费

EEGSIGNALPROCESSING第一页，共二十七页。EEGsignalmodelling1Availablefeatures2Classificationalgorithms3IndependentComponentAnalysis4ContentSparseRepresentation5第二页，共二十七页。1EEGsignalmodellingBioelectricity1Signalgenerationsystem2第三页，共二十七页。bioelectricitySignalgenerationsystemExcitationmodel第四页，共二十七页。signalgenerationsystembioelectricityLinearModel第五页，共二十七页。signalgenerationsystembioelectricityNonlinearModel第六页，共二十七页。2AvailablefeaturesBasicfeatures1Modernmethods2第七页，共二十七页。TemporalAnalysisSignalSegmentation:labeltheEEGsignalsbysegmentsofsimilarcharacteristics.basicfeaturesModernmethods第八页，共二十七页。TemporalCriteriabasicfeaturesModernmethods第九页，共二十七页。FrequencyAnalysisSuboptimalDFT,DCT,DWT;OptimalKLT(Karhunen-Loève)Demerits:completestatisticalinformation,nofastcalculation.basicfeaturesModernmethods第十页，共二十七页。SignalParameterEstimationARmodel:Merits:OutperformDFTinfrequencyaccuracy.Demerits:sufferfrompoorestimationofparameters.Improvements:accurateorder&coefficients.modernmethodsBasicfeatures第十一页，共二十七页。ARcoefficientsestimationmethodsYule-Walkeraryule(x,p)Merits:ToeplitzmatrixLevinson-Durbin,fastest!!!Demerits:withwindowbadresolutionofPSDmodernmethodsBasicfeatures第十二页，共二十七页。ARcoefficientsestimationmethodsCovariancemethodarcov(x,p),armcov(x,p)Merits:withoutwindowgoodresolutionofPSDDemerits:slowBurgarburg(x,p)Merits:accurateapproximationofPSDDemerits:lineskewing&splittingmodernmethodsBasicfeatures第十三页，共二十七页。modernmethodsBasicfeaturesComparison第十四页，共二十七页。PrincipalComponentAnalysisUsesameconceptasSVDDecomposedataintouncorrelatedorthogonalcomponentsAutocorrelationmatrixisdiagonalizedEacheigenvectorrepresentsaprincipalcomponentApplicationdecomposition,classification,filtering,denoising,whitening.modernmethodsBasicfeatures第十五页，共二十七页。3SparseRepresentationSparseApproximation1SparseDecomposition2第十六页，共二十七页。Over-completedictionaryatomsHilbertspace:Signal:Error:“Sparse〞:l<