1、S1 21.2.,E-mail: 210046210003E-mail: : SFourierS SS SFourier:SEEG time-frequency analysis based on the improvedS-transformZHANG Shaobai1,HUANG Dandan21. Computer Department ,Nanjing University of Posts and Telecommunications ,nanjing,210046E-mail: 2. Computer Department ,Nanjing University of Posts
2、and Telecommunications ,nanjing,210003E-mail: Abstract: S-transform, which is a combination of short-time Fourier transform and wavelet transform, has attract intensive interest inrecent years as an important tool to investigate non-stationary signal time-frequency distribution. S transform can be s
3、elf-improved bythe EEG characteristics to select a suitable mother wavelet. The improved S-transform will be used to analyze the time-frequency ofthe EEG characters. A comparison among the Short-time Fourier transform, wavelet transformation and the improved S-transformindicates that improved S-tran
4、sform gives the best energy distribution in the time-frequency filed.Key Words: S transform; Time-frequency analysis; Electroencephalography(EEG); Wavelet Transform1EEG EEG2 SSCohenSt SS SP(t, )2.1Stockwell R GSTFTSxt L2 RS wt,dSd*S61073515. WT , d =-xtx tw t - ,d dt 1978-1-4673-5534-6/13/$31.00 c 2
5、013 IEEE3410t-e dt n mST , f =e W , d ,eN t X ( f ), jTfnn 1 N1 n mX( )e 2 mS ( jT , )X ( )G ,e-i 2 mj1,0 t 1t = xt -x dt =tt - 2-tSST , f =- x tf2 e-22f 2-i 2 tS 2342 X (3NTX ()mNT)i 2 f 3 Hn mNT MTf d =f -1 5 G n mNT MTw t,f =f2e-t 2 f 22e-i 2 tS4 6 SnNT, jT345SS t -Fourierff SSnNT, jTS FFTi 2 fS3
6、 SSTFTSfSSw t,dSSS2.2 S 3.1x(t)S x(kT ) Mayer Morlet HaarDaubechiesSymletsk =0,1,2 N 1 T2 STNT k =0nNTN1X ( ) x(kT )e-i 2 nk NNT n m=0 NTS1 x(kT )mNT2n mNT MT2 2n2 NnnS10(5)FFTBSn B n tt0,10B10B SB f t =A f tf1 tf - 2-tf 2 tfA3.2 St 0 t 10,B678S S S2013 25th Chinese Control and Decision Conference (
7、CCDC)3411B f t eS,f = xt Aft- f1t- f-2-t- f2t- fe dt -2tS S WVD-i 2 ftB f t e - i 2 ft =A f tf1 tf - 2-tf 2 tf e - i 2 ft9STFTHeisenbergSTFTx t S8S-i 103SSTFTB SfSTFTA f4B S SBGabor 3 STTF11025HZ 2001,2068 680141 42 SS2S3STFT3 4 STFT2 4 SS0.01s 0.02s 0.03s 0.04s800HZ-1200HZSTFT S0.2814 0.5312 2.5321
8、341222SSS2013 25th Chinese Control and Decision Conference (CCDC)S5 5Ventosa S Simon C.The S-transform from a wavelet point ofviewJ.IEEE TransactionsS B on Signal Processing 2008 56(7):2771-2780.SSSTFT62007.S- D .(MCE)7. S2011 37(2):217-219.J.S 8 . B J.2011 28(11):214-254.9 . S2006 26(2) 28-30.J.123
9、4J 2004 8(1) 152-154.Schroder M Bogdan M,Hinterberger T et al. AutomatedEEG feature selection for brain computerinterfacesC/Proceedings of First International IEEEEMBS Conference on NeuralEngineering.S.I:IEEE,2003:626-629.J 2012,38(1) 21-37.SJ 2009 30(2) 305-310.1011121314J.J.D .32(4):77-84. QRS B. BB. 2011.Hermite J.2011 28(1):80-87.J. 20122012,35(4):88-93.2012,29(2) 184-187. S2013 25th Chinese Control and Decision Conference (CCDC)3413
