A new method for person identification in a biometric security system based on brain EEG signal processing

Recently, researches in a biometric security tend to use new types of biometric that based on physiological signals, such as EEG and ECG signals, rather than more traditional biological traits. Since it is very hard to fake an EEG signature or to attack an EEG biometric system, this paper presented a biometric security system that based on EEG signal processing. A new investigated method for person identification using the EEG Brain Signal processing is introduced. The proposed method based on executing a voting scheme between the 3 feature extraction methods which achieved maximum classification rates in the preliminary test. Preliminary test used Discrete Fourier Transform (DFT) and Wavelet packet decomposition (WPD) for features extraction with two different measures with each of them, thus a total of 4 different methods, produced 4 different features sets. Classification rates were 93%, 87% & 93% using the 3 recommended features sets. After executing the proposed voting scheme, classification rate increased to 100%, for 3 subjects´ experiment which surpassed the results from the previous works in this application. Multi-layer Perceptron Neural Network trained by a standard back propagation algorithm is used as a classifier. Taking into account of reducing distraction to subjects, 4 channels only were used in the experiment and the subject need only to sit with eyes closed and quiet, which free the physical requirements of users and the condition of applying environment.