AnovelapproachframeworkbasedonstatisticsforreconstructionandheartrateestimationfromPPGwithheavymotionartifacts
摘要: Oneofthemostimportantapplicationsofphotoplethysmography(PPG)signalisheartrate(HR)estimation.Foritsapplicationsinwearabledevices,motionartifact(MA)maybethemostseriouschallengeforrandomnessbothinformatandtemporaldistribution.Thispaperproposesanadvancedtime-frequencyanalysisframeworkbasedonempiricalmodedecomposition(EMD)toselectspecifictimeslicesforsignalreconstruction.Thisframeworkoperateswithatypeofpre-processingcalledvariancecharacterizationseries(VCS),EMD,singularvaluedecomposition(SVD),andapreciseandadaptive2-Dfiltrationreportedfirst.ThisfiltrationisbasedonHarrwavelettransform(HWT)and3rdordercumulantanalysis,tomakeithaveresolutioninboththetimedomainanddifferentcomponents.Thesimulationresultsshowthattheproposedmethodgains1.07inabsoluteaverageerror(AAE)and1.87instandarddeviation(SD);AAEs1stand3rdquartilesare0.12and1.41,respectively.ThisframeworkistestedbythePhysioBankMIMICIIwaveformdatabase. ...
(共12頁)
開通會員,享受整站包年服務