2022年医学专题—EEG-signal-processing-脑电信号处理方法算法.pptVIP免费

EEGSIGNALPROCESSING第一页，共二十七页。EEGsignalmodelling1Availablefeatures2Classificationalgorithms3IndependentComponentAnalysis4CONTENTSparseRepresentation5第二页，共二十七页。1EEGsignalmodellingEEGsignalmodellingBioelectricity1Signalgenerationsystem2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。2AvailablefeaturesAvailablefeaturesBasicfeatures1Modernmethods2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。3SparseRepresenta...