Abstract
This study has simulated an ECG-based person identification system. This research is a development of previous studies in which accuracy improvement is the main focus to be achieved. One lead ECG signal from 11 subjects was simulated in this work. ECG signals from each person are then segmented into 10 windows to become training data and test data. Variational Mode Decomposition (VMD) and Ensemble Empirical Mode Decomposition (EEMD) methods are used to decompose ECG signals into 5 sub signals. Feature extraction based on statistical calculations was applied at each level of decomposition to obtain the characteristics of the ECG signal. Mean, variance, skewness, kurtosis and entropy have been evaluated as predictors. Support vector machines and 10-fold cross validation are used to validate the performance of the proposed method. Our simulations demonstrate that the proposed method outperforms several previous studies and achieves an accuracy of up to 98.2%.
Keywords
References
- Autori zadržavaju autorska prava i pružaju časopisu pravo prvog objavljivanja rada i licenciraju ga Creative Commons licencom koja omogućava drugima da dele rad uz uslov navođenja autorstva i izvornog objavljivanja u ovom časopisu.
- Autori mogu izraditi zasebne, ugovorne aranžmane za neekskluzivnu distribuciju rada objavljenog u časopisu (npr. postavljanje u institucionalni repozitorijum ili objavljivanje u knjizi), uz navođenje da je rad izvorno objavljen u ovom časopisu.
- Autorima je dozvoljeno i podstiču se da postave objavljeni rad onlajn (npr. u institucionalnom repozitorijumu ili na svojim internet stranicama) pre i tokom postupka prijave priloga, s obzirom da takav postupak može voditi produktivnoj razmeni ideja i ranijoj i većoj citiranosti objavljenog rada (up. Efekat otvorenog pristupa).