(For USM Staff/Student Only)

EngLib USM > Ω School of Electrical & Electronic Engineering >

A comparative study of different types of mother wavelets for heartbeat biometric verification system / Lee Jian Han

A comparative study of different types of mother wavelets for heartbeat biometric verification system_ Lee Jian Han_E3_2017_MFAR
Kebelakangan ini, teknologi biometrik terkini sedang beralih arah kepada penggunaan isyarat elektrokardiogram (ECG) sebagai satu modaliti terbaru untuk sistem pengesahan. Isyarat ECG mengandungi maklumat yang mencukupi untuk mengesahkan individu kerana ia adalah unik kepada setiap individu. Satu daripada teknik yang dipercayai yang mampu mengektrak maklumat penting daripada isyarat ECG ialah dengan menggunakan transformasi wavelet. Namun, terdapat satu cabaran dalam mengimplimentasikannya kerana perbezaan jenis dan peringkat ibu wavelet akan mempengaruhi prestasi pengesahan. Jadi, dalam kajian ini, satu kajian perbandingan akan dibuat bagi menyemak jenis dan peringkat ibu wavelet yang optimum yang boleh mewakili ciri yang terbaik untuk sistem verifikasi ini. Tiga jenis ibu wavelets iaitu Symlet, Daubechies dan Coiflet dengan turutan peringkat dari satu kelima telah dikaji di dalam kajian ini. Ciri yang diektrak kemudiannya dilatih dengan pengelas Mesin Vektor Sokongan (SVM) bagi membina model untuk mengesahkan ciri yang diektrak. Prestasi sistem pengesahan biometrik ECG ini dinilai dengan Penerima Operasi Ciri (ROC) dan Kadar Ralat Sama (EER). Keputusan kajian menunjukkan sistem yang telah dibangunkan ini menghasilkan prestasi yang terbaik apabila ibu wavelet Coiflet peringkat ketiga digunakan sebagai ciri dengan prestasi EER bernilai 10.755% diperolehi. Recently, advanced biometric technology is turning to the use of electrocardiograms (ECG) signal as new modality for verification system. The ECG signal contains sufficient information to verify an individual as it is unique to everyone. One of the feasible methods to extract the salient information from ECG signal for verification is by using wavelet transform. However, there is a challenge in implementing it as different types and orders of mother wavelet used will yield different verification performance. Therefore, in this study, a comparative study is done so as to investigate the optimum type and order of mother wavelet that represents the best feature for the verification system. Three different types of mother wavelets i.e. Symlet, Daubechies and Coiflet with order ranging from one to five have been studied in this research. The extracted features are then trained by using SVM classifier to generate a model to verify the features. The performance of the ECG biometric verification system is evaluated with the Receiver Operating Characteristic (ROC) plot and Equal Error Rate (EER). Experimental result showed that the developed system achieves the best performance when the 3rd order Coiflet is used as feature with an EER score of 10.755% is achieved.
Contributor(s):
Lee, Jian Han - Author
Primary Item Type:
Thesis
Language:
English
Subject Keywords:
The ECG signal; Symlet, daubechies and coiflet; receiver operating characteristic (ROC); Equal error rate (EER)
Sponsor - Description:
Pusat Pengajian Kejuruteraan Elektrik & Elektronik -
First presented to the public:
1/1/2017
Original Publication Date:
12/3/2018
Previously Published By:
Universiti Sains Malaysia
Place Of Publication:
School of Electrical & Electronic Engineering
Citation:
Extents:
Number of Pages - 79
License Grantor / Date Granted:
  / ( View License )
Date Deposited
2018-03-12 11:22:05.131
Date Last Updated
2020-05-28 18:14:56.59
Submitter:
Mohd Fadli Abd. Rahman

All Versions

Thumbnail Name Version Created Date
A comparative study of different types of mother wavelets for heartbeat biometric verification system / Lee Jian Han1 2018-03-12 11:22:05.131