32! 4" 2012 04# $ JournalofXi antechnologicaluniversity Vol.32No.4 Apr.2012 : 1673 9965(2012)04 310 05 ( # Z[? @," A,BC (,,710032) : <Q [ "7#:,7#+, G/ R, ($ " % I % G/ "7#. (bior3.7$ Malat ECG7#G/A, % $ 5@ 7#G/, <$ " % +,. DE $ %]Q [ "7 #:, "7# Q78, $ " % ] "7#: +,. : "7#;$ "; % ;+, R : TP391 : A 8,- \,-, C`HI,)`0.05~100Hz, P\ 7K / 6 &'D I_.8,- 6 A :1G J,) H1Hz,?`X 0 ;2K6,)N, 5Hz ~2000Hz;,?` 0 ; 3)6, )#50Hz 60Hz. 8,- b_slm: A,-$F, K$F Q) #, _ ) / H H Q), UFT _E. [1] G W bno Obior3.7_`F,DEF C$F 8,- b_,-, A b _. LM:, [2] 6%C $F F X!T LM:, LM:. 1 / ' 1.1 _ G<@=0> 0 F [3] (Discrete WaveletTrans? form,dwt), F F Ψ 犪, 犫 ( 狋 ) 犪, 犫 0,O 犪 = 犪犼 0, 犫 = 犽犫 0 犪犼 0, 犼, 犽 犣, F - 犼 /2 Ψ 犼, 犽 ( - 犼狋 )= 犪 0 Ψ( 犪 0 狋 - 犽犫 0), 犼, 犽 犣 (1) 犳 ( 狋 ) 犔 2 ( 犚 ), 犔 2 ( 犚 ) a YG L, F - 犼 /2 犠犳 ( 犪, 犫 )= 犪 F L 0 + 犳 ( 狋 )= 1 犆 Ψ + - - + 犳 ( 狋 ) 珡 Ψ 犼, 犽 (- 犼犪 0 狋 - 犽犫 0)d 狋 (2) 1 `F M. 犆 Ψ 1.2 =0> 0 Malat - 犠犳 ( 犪, 犫 )Ψ 犼, 犽 ( 狋 )d 犪 d 犫 (3) > N >? (MultResolution Analysis, MRA) A >?, G E_ F G L,- D b >?, D OAE,. Malat$F [4] F > S Y $F,[ $F, 犳 ( 犽 )`,- 犳 ( 狋 ) P, 犳 ( 犽 )= 犮 0, 犽,,- 犳 ( 狋 ) _ F >!:2011 12 08 &' :O (1955?),P, +, 78 123 7K.E?mail:nnyy930@sohu.com.
6" O R:8,- F,-$F] 311 犮犼 +1, 犽 = 狀 犣犮犼, 狀犺狀 -2 犽, 犼, 犽 犣 犱犼 +1, 犽 = 狀 犣犮犼, 狀犵狀 -2 犽, 犼, 犽 犣 (4) : 狕 `^ 6; 犮犼, 犽 ` ; 犱犼, 犽 `F ; 犺, 犵 > S K; 犼 ` L.F Y F > L=$,D Y 犮犼, 犽 = 犮犼 +1, 狀犺犽 -2 狀 + +1, 狀犵犽 -2 狀, 犼, 犽 犣 (5) 狀 犣狀 犣犱犼 1.3 _?;,,- O F,- c, F O. OF 3 :1 \ : M" )Z H, F A CZ H+ ;2 A :,-,- A _E; 3; : M A _E; 4_ :,- Y] c > AE. F Xc S >?, # Obior3.7F G 8,- > S Y. 2 Q 2.1 _ @,-b_a. ],- E >, Z,- AE >.Q,- ;_ >`_ [4] :1Q,- F >. F > VW, > # $;2F > ) C 0: > ) C C 0,-, cae,,b _ G>;3QF Y:[ F > N V ) S V ) QF Y. PE MIT/BIH ArrhythmiaDatabase,P 360Hz.Bior3.7F )'D 1. 2.2 AB@-BC" D 1 G J >, A AE8,-, E,-PEbior3.7F 9V>, d 9,- Z b_,-, d 9 F +, ; G J. 1 d 1 ~d 9 R. 1 E, Tab.1 Thepassbandofsignal sscales 5 /Hz d 1 90~180 d 2 45~90 d 3 22.5~45 d 4 11.25~22.5 d 5 5.625~11.25 d 6 2.8125~5.625 d 7 1.40625~2.8125 d 8 0.703125~1.40625 0.3515625~0.703125 d 9 2.3 -BC F )6 50Hz] QG, 8,- )5 a. 1 ": d 2 >, PE CF 50Hz D F ],U ; _. C 烄 sgn(cd( 犼 )) ( 狘 cd( 犼 ) 狘 -th( 犼 ), cdd( 犼 )= 烅 狘 cd( 犼 ) 狘 th( 犼 ) (6) 烆 0, 狘 cd( 犼 ) 狘 <th( 犼 ) :cd( 犼 )` F ;th( 犼 )`' D C, #O th( 犼 )=δ 犼 ( 2 log( 犔犼 )) 1/2, 犼 =1,2,3 (7) log2 :δ 犼 ` _ ; 犔犼 ` P :,cdd( 犼 )` C,- F. 2.4 G 4D -BCHb F HK6, H )> N, AE ECG,-? ab,`" ab AE,- _, PE CF>,-. d 1 ~d 3 PE CF [3] U ; )_. d 4 ~d 8 F PE C $F. C cdd( 犼 )= 烄 sgn(cd( 犼 )) ( 狘 cd( 犼 ) 狘 -0.1 犼 th( 犼 )), 烅 狘 cd( 犼 ) 狘 th( 犼 ) (8) 烆 0, 狘 cd( 犼 ) 狘 <th( 犼 ) 3 Q# PE MIT/BIH ArrhythmiaDatabase [5] Q $F b, H 6,`" R$F,- c,,- bi H _,E)`0.2Hz _,-
312 $ 31! HG J,)`50Hz _,- H) 6,E 0_ HK6. 124Q b_ $F Xc* 1 R, *1(a) 6* G J A b,8,- AE, c.*1(b)a"* \ 50Hz )6 SK6 A b. 1 I @J B@L, B@KL, L23 Fig.1 Comparisonoforiginalsignal,noisesignal,denoisesignalandspectrumofbeforeandafterdenoising 4 > 8,- [6] A P QRS T U PR " QT " ST, #$ W *. 4.1 M - LM: [7] QRS P T, PE F 6%C X! $F LM: [8],LM: > 1)QRS R :1 8,- 8 F >, O Y,-,- > ;2 >, C:, CN :;3 N C: C(tr1), C: 0.6tr1,H C UR` R %C :..E a?" R. [ QRS L:,Q S R G, R G > S 50ms NF C,U Q S. 2)P T HP T C F, _ 6, QRS, QRS G.P T :1 P P : QRS, E R % C: P,E R R %C: 0.8RR " M :, QRS R %C: OQ:;2 P P, E%C : Q E,- C:,#$ : R G C, C H C 犜 p ( 犜 plmc [ _ P CN b, P D[ P C, a W C), R` :` P %C :;3 T P : R %C: QRS R %C: M, R %C:0.5RR " X ;4 T P, -, E%C: Q E,- C:, : R G C C/, :` T %C:.
6" O R:8,- F,-$F] 313 4.2 - N 124Q LM: $F X c*2 R. *2 " QRS P S T. 2 M Fig.2 Featurepointscalibration 5 @ %& PE,_ S b_$f >?, #$LM: ] LM: $F >?.,_ 熿 SNR=10log 燀 狀 犡 2 ( 狀 ) 狀 燄 [ 犡 ( 狀 )- 犡 2 ( 狀 )] 燅 : 犡 ( 狀 )`,-; 犡 ( 狀 )`b_,-. #$ (9) RMSE= 1 狀 [ 犡 ( 狀 )- 犡 ( 狀 )] 2 (10) 槡狀,_ -, -F, b_,- -,-,b_ cu-. 2 Q8 >?. Q b ", $F 8 _,- LM: 100%, 8 S, T CS R C, T GR, GR ] L M:. 208 214 221 231 8 S,-, LM: R T. 2 -,; 0 Tab.2 Thedigitalanalysisofalgorithminthepaper Cb_ LM: `- / / /%,_ /mv QRS P T QRS P T QRS P T 208 20.8709 0.1816 3 3 3 0 0 0 23.08 23.08 23.08 214 27.1670 0.1260 2 2 2 0 0 0 14.29 14.29 14.29 217 25.2403 0.1695 0 0 0 0 0 0 0.00 0.00 0.00 221 12.2301 0.1996 5 5 5 0 0 0 33.33 33.33 33.33 231 18.2029 0.1838 2 2 2 0 0 0 13.33 13.33 13.33 108 17.6282 0.1866 0 0 0 0 0 0 0.00 0.00 0.00 234 12.3983 0.1966 0 0 0 0 0 0 0.00 0.00 0.00 124 15.4150 0.1903 0 0 0 0 0 0 0.00 0.00 0.00 123 18.0145 0.1849 0 0 0 0 0 0 0.00 0.00 0.00 121 17.5751 0.1854 0 0 0 0 0 0 0.00 0.00 0.00 C 18.4742 0.1804 8.40 8.40 8.40 3` A $F [2] MIT/BIH _ b SLM: Xc>?, E E H $F. 2~3, $F,_18.4742, 0.1804, A$ F,_, F, E $F 3 O, -;,-b_. LM:, E$ Tab.3 Thealgorithmdataexisited F QRS P T 8.403,QRS P.TD$F F C FC$F A$F, T /%,_ /mv H A$F, $F H8 S QRS P T T C R C 8,-LM: 14.0967 0.4519 9.0725 22 7.2 9:, T R.
314 $ 31! 6 ' PEF Cb_$F,- b_, LM:, Xc E$F PEF 6%C X! $F,- c8,-aelm, A b,- _, EG LM:. LM:#$ C *TW A, [ C W *-,. < : [1] IU,VW. 6,-,-S [M]. X: X&2 Y 3!,1997. YANG Fu?sheng,LV Yang?sheng.Processingand PaternRecognitionofBiomedicalSignal[M].Tian jin:tianjinscienceandtechnologytranslationpub lishingcompany,1997.(inchinese) [2] I.8,-,-S $F78[D]. :,2009. NIUChuan?li.StudyonthePretreatmentandWave formdetectionalgorithmofecg[d].beijing:beijing JiaotongUniversity,2009.(inChinese) [3] IU.F >? DE[M]. :& 34,1999. YANG Fu?sheng.The Engineering Analysisand ApplicationofWaveletTransform[M].Beijing:Sci encepress,1999.(inchinese) [4].MATLABF >?[M]. :% 34,2009. ZHANG De?feng.WaveletAnalysison MATLAB [M].Beijing:MachineryIndustryPress,2009. (inchinese) [5]. MIT?BIH 8,-[J].&2 >,2010(9):27. SONG Chun?li.HowtoReadtheECG Signalsof MIT?BIH[J].ScienceandTechnologyInformation, 2010(9):27.(inChinese) [6] Z, ), Z,.ECG8[>U LMN O 78[J]. 6 [,2007,11(5): 344. HUANG Wen?xia,CHENJian?hua,HUANG Bo? qiang,etal.thefeatureextractionresearchofecg HeartbeatClassification[J].BiomedicalEngineering andclinic,2007,11(5):344.(inchinese) [7] LICui?wei,ZHENG Chong?xun,TAIChang?feng. DetectionofECGCharacteristicPointsUsing Wave lettransform[j].ieee TransactionsonBiomedical Engineering,1995,42(1):20. [8] SANXENASC,KUMAR V,HAMDEST.Feature Extractionfrom ECGSignalsUsing WaveletTrans formsfordiseasediagnostics[j].internationaljour nalofsystemsscience,2002,33(13):1073. 犘狉狅犮犲狊狊犻狀犵犃犾犵狅狉犻狋犺犿犅犪狊犲犱狅狀犠犪狏犲犾犲狋犜狉犪狀狊犳狅狉犿犪狀犱犛犻犿狌犾犪狋犻狅狀犳狅狉犈犆犌犛犻犵狀犪犾 犖犐犢狌犪狀, 犠犃犖犌犡犻犪狅? 犾犻, 犔犝犠犲狀? 狕狅狀犵 (SchoolofElectronicInformationEngineering,Xi antechnologicaluniversity,xi an710032,china) 犃犫狊狋狉犪犮狋 : Forefectivelyeliminatingnoiseofsignalandaccuratelydemarcatingcharacteristicpointsof signal,denoisingalgorithm ofwavelettransform withthresholdandalgorithm oftimedomainpeak orientationareadoptedtoprocessecgsignal.thebior3.7waveletisusedtodecomposeecgsignal accordingtomalatalgorithm,thenalgorithmforwaveletreconstructionwithcombinationofsoftand hardthresholdisusedtodenoise,finalythealgorithmoftimedomainpeakorientationbasedonwavelet transformisusedtodemarcatefeaturepoints.simulationresultsshow thatthenoiseisremoved efectively,theecgsignalisretained,andfeaturepointsareaccuratelydemarcated. 犓犲狔狑狅狉犱狊 : ECGsignal;wavelettransform;thresholddenoise;demarcationoffeaturepoints (STUV W XY7)