Automation&Instrumentation2012(12):1001-9944(2012)12-0046-04徐寒,王冬冬,蒋同斌(,223003):在采集心电信号数据的过程中,必然会涉及到肌电干扰、基线漂移和50Hz工频干扰,而使用常规系统辨识法则常常在一定程度上难以鉴定心电信号的特性。中值滤波器是一种操作简单的、高速的非线性信号滤波器,它常用于心电信号中低频去噪过程,如基线漂移。因为WTS的二进小波是一组带通滤波器,不同尺度有不同的频带,小波变换被选定用来分解原始信号,小波变系数的重建形成了消除干扰的心电信号。采用模拟实验是要确定如何进行自适应的阈值选取,适当的分解层数和小波函数。通过使用MIT/BIH数据库的心电信号,并结合计算机仿真形成的心电信号来对该方法进行检验。结论表明此算法可有效抑制心电信号中的主要噪声,满足心电波形临床分析和诊断的需求。:;;;:TP13:BResearchontheECGSignalDenoisingAlgorithmBasedonWaveletTransformandtheMedianFilterXUHan,WangDong-dong,JIANGTong-bin(FacultyofMathematicsandPhysics,HuaiyinInstituteofTechnology,Huaian223003,China)Abstract:IntheprocessofECGsignaldataacquisition,itwasnecessarytoinvolvetheelectromyographicinterfer-ence,baselinedriftand50Hzinterference,whileusingtheconventionalsystemidentificationrulesoftentosomeex-tentandwasdifficulttoidentifythecharacteristicsofECGsignal.Medianfilterwasakindofsimpleoperation,highspeednonlinearsignalfilter,itwascommonlyusedintheECGsignaldenoisingprocessinlowfrequency,suchasbaselinedrift.SinceWTStwodyadicwaveletwasasetofband-passfilterswithdifferentscales,differentfrequencybands.Thewavelettransformwasusedtodecomposetheoriginalsignalselected.WavelettransformcoefficientsofthereconstructionwasformedtoeliminateinterferenceinECGsignal.Thesimulationexperimenttodeterminehowadaptivethresholdselection,appropriateleveldecompositionandwaveletfunction.ThroughusingtheMIT/BIHdatabaseofECGsignal,andcombiningwiththecomputersimulationoftheformationofECGsignalonthemethodoftesting.ConclusionshowsthatthisalgorithmcaneffectivelyrestrainthemainnoiseinECGsignals,meettheECGwaveformanalysisofclinicalanddiagnosticrequirements.Keywords:ECG;denoisingalgorithm;wavelettransform;medianfilter:2012-06-19;:2012-10-22:(1962—),,,,、。、。,[1]。,、,。,,,,;462012(12)。,,。,[2-4]。J.Morlet1984,、、、DonohoJohnstone[5],,QS,。,。,。[6],。。[7],,。,。[8],。,。MIT/BIH。。1,1Hz,;,。1.1,,。,。,,。,。。、。,()。,。,。,,。,。,。,,。1.2(),,。,,,、()。,。:x={x(1),x(2),x(L)}L,L。Xx={x(-k+1),x(-k+2),…,x(k+L)}x(n)=x(1)(-k+1)≤n≤1x(n)1<n<Lx(L)L≤n≤(L+k≤≤≤≤≤)(1)n,1≤n≤L,x(1)(n),2k+1:x(n-k),x(n-k+1),…x(n),…,x(n+k-1),x(n+k)。,x={x(1)(1),x(1)(2),…x(1)(L)}X2k+1;x(1)2k+1,x(2),x(p)={x(p)(1),x(p)(2),…,x(p)(L)}2k+1XP47Automation&Instrumentation2012(12),。,QRS,TP,,。,(QRS,TP),。1.3,30min105MIT-BIH,1。1(b),m=2,360Hz;0.5Hz。1(c)。,。,。,,。,。,,。,。,,,,,。250Hz,5Hz~2kHz,。,。,。,,,、、。,。,,,。,,。2.1,,,,。,Donoho。Tf赞,f赞h,f赞s,:f赞h=f,f>T0,f≤≤T(2)f赞s=sign(f)f-≤≤Tf>T0f≤≤T(3),,,。,:①,Stein():T,,T,,。1Fig.1Lowfrequencyinterferencesignalfilterthesimulationresults(a)(b)(c)(d)482012(12)②sqrt(2log(length(X))),,sqrt(2log(length(X)))。③:。(SNR),。④-:-,。,,。。2.250Hz,。,。。,,。,。3,MIT/BIH。10k,。。,。,。2,。2,,。,,QRS。4,,,。,。:[1],.—-:[J].,2010,29(5):11-16.[2],.[J].,2002,15(1):64-67.[3]AKZiarani,AKonrad.Anonlinearadaptivemethodofelimi-nationofpowerlineinterferenceinECGsignals[J].IEEETransactionsonBiomedicalEngineering,2002,49(6):540-547.[4],.[J].,2006(8):901-905.[5]DavidLDonoho,IainMJohnstone.Idealspatialadaptationbywaveletshrinkage[J].Biometrika,1994,81(3):425-455.[6],.[J].,2002,24(2):110-117.[7],.ECG[J].,2004,17(3):832-841.[8],,,.[J].,2007,28(1):201-206.[9]JAVanAlste,TSSchiler.Removalofbase-linewanderpower-lineinterferencefromECGbyanefficientFIRfilterwithare-ducednumberoftaps[J].IEEEBME,1985,32(12):1052-1060.■2Fig.2FilteringalgorithmforremovalofwhitenoiseeffectchartV/mV420-200.81.62.43.24.04.85.66.47.28.0t/s(a)V/mV420-200.81.62.43.24.04.85.66.47.28.0t/s(b)2013《》邮发代号:6―20定价:8.00元/期■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■49