316 208

Cited 0 times in

위전도 신호처리를 위한 웨이브렛 필터와 디지털 필터의 비교

Other Titles
 Comparison of digital filters with wavelet multiresoludion filter for electrgastrogram 
Authors
 유창용  ;  남기창  ;  김수찬  ;  김덕원 
Citation
 Journal of Biomedical Engineering Research (의공학회지), Vol.23(2) : 109-117, 2002 
Journal Title
 Journal of Biomedical Engineering Research (의공학회지) 
ISSN
 1229-0807 
Issue Date
2002
Keywords
Electrogastrography ; FIR filter ; IIR filter ; wavelet ; RSE ; SNR
Abstract
Electrlgastrography(EGG) is an noninvasive method for measuring gastric electrical activity on the abdomen resulting from muscle. EGG signals have a very low frequency range(0.0083~0.15㎐) and extremely low amplitude(10~100μV). consequently, EGG signals is easily influenced by other noises. Both finite impulse response(FIR) and infinite impulse response (IIR) filters need high orders or have phase distortions for passing very narrow bandwidth of the EGG signal. In this study, we decomposed EGG signals using a wavelet multiresolution method with Doubechies mother wavelet. The EGG signals were decomposed to seven levels. We reconstructed signal by summing the decomposed signals from level four to seven. To evaluate the performance of the wavelet multiresolution filter(WMF) with simulated EGG signal using two kinds of FIR and four kinds of IIR filters., we used two indices; signal to noise ratio(SNR) and reconstruction squared error(RSE). The SNR of WMF had 9.5, 6.9 and 4.7 dB bigger than that of the other filters at different noise levels, respectively. Also, The RSE of WMF had 1.22×106, and 1.22×106 smaller than that of the other filters at different noise levels, respectively. The WMF performed better in the SNR and RSE than two kinds of FIR and four kinds of IIR filters.
Files in This Item:
T200210932.pdf Download
Appears in Collections:
1. College of Medicine (의과대학) > Dept. of Medical Engineering (의학공학교실) > 1. Journal Papers
Yonsei Authors
Kim, Deok Won(김덕원)
URI
https://ir.ymlib.yonsei.ac.kr/handle/22282913/144651
사서에게 알리기
  feedback

qrcode

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.

Browse

Links