International Journal of Science and Research (IJSR) ISSN (Online) : 2319-7064 Index Copernicus Value (2015) : 78.96 | Impact Factor (2015) : 6.391 Volume 6 Issue 9, September 2017 www.ijsr.net Licensed Under Creative Commons Attribution CC BY EEG Based User Identification and Verification Using the Energy of Sliced DFT Spectra Hend A. Hadi 1 , Dr. Loay E. George 2 1, 2 College of Science, Computer Science Department, Baghdad University, Baghdad, Iraq Abstract: Electroencephalogram signals reflect the electrical activity of the brain; EEG signal is the measurement of voltage fluctuations coming from ionic stream within the neurons of the brain. They have been explored in medical researches to diagnose some brain diseases such as Alzheimer's and epilepsy, and have been used in Brian computer interface (BCI) applications. Recently EEG signals are being investigated for identification and verification applications because they show evidence against falsification or replication since the brain activity of people is distinctive. In this paper a promising EEG-based identification and verification system is presented. A feature set based on the energy distribution of Fourier accumulative components is proposed, and some Euclidean distance measures are used for matching. This system was tested on the EEG public CSU dataset which was collected from 7 healthy volunteers. The attained identification results are encouraging with best recognition result is (100%), the tested feature sets were extracted under the condition "they should extract from single task & signal channel". The verification results indicated that the minimum achieved HETR is (0.4%), these results are considered competitive when compared with the results of other recently published works. The adopted condition "one channel per single task" was aid to achieve less computational complexity and, consequently, little execution time is required. Keywords: EEG signal processing, Discrete Fourier transform, Energy features 1. Introduction EEG signals draw the attention of researchers because they can lead to distinctive features about the user identity. Also, they are robust against falsification or replication. Other Biometrics such as fingerprint, hand geometry, facial features, and voice characteristics can be forged using spoof attack. Due to the development of biomedical instrumentation these signals are acquired easily using portable devices with dry electrodes. EEG signals are measured with the electrodes placed on different places of the scalp (Abo-Zahhad, Ahmed and Abbas 2015), (Rodrigues, et al. 2016).The first proposed works that studied the EEG signals as biometric was by Poulos et al.(Poulos, et al. 1999), (Poulos, Rangoussi and Alexandris 1999), (M. Poulos, et al. 1999).Then, this research area had received big interests due to its potential in biometrics systems. (Palaniappan 2006) proposed the use of AR coefficients, channel spectral powers, differences of inter-hemispheric channel spectral power, inter-hemispheric channel linear and non-linear complexity for feature extraction after filtering the signals using Finite Impulse Response (FIR) filter, and then he reduced the feature vector size using Principal Component Analysis (PCA), also he used Linear Discriminant Classifier (LDC) to classify5 subjects and achieved recognition rate up to (100%) with features combined from Rotation, Math., Letter, Baseline tasks. (Palaniappan 2008) Proposed same system in (Palaniappan 2006) for subject authentication and achieved best accuracy with FAR and FRR both zero. (Kumari and Vaish 2016) Proposed a system based on the fusion of features from different mental tasks using Canonical Correlation Analysis (CCA). They used Empirical Mode Decomposition (EMD) and information theoretic measure with statistical measurement for feature extraction, then classified the 7 subjects using Linear Vector Quantization (LVQ) neural network and its extension LVQ2;they achieved best recognition rate (96.05%). (Bajwa and Dantu 2016) Proposed to use EEG signals for both authentication and cryptographic key generation; they used Fast Fourier Transform (FFT) to decompose the signal into frequencies that make it up, and then they used Daubechies (Daub8) to break the signal into to obtain the five major rhythms that composes the brain signal, then they compute the energy of each sub band to obtain the relevant features. Two type of classifiers Support Vector Machine (SVM) and Bayesian network were used, and the best achieved accuracy was (100%). While in this work the DFT spectra of the signal is partitioned into slices, then the average energy of each slice is determined to obtain the relevant features without the need for wavelet decomposition step. However the researches on EEG-based recognition have faced complications in feature extraction and their combination or in the fusion steps to select the best features for classifiers, they used features extracted from different channels and tasks, and then they tried to use features fusion to generate final feature vector. EEG-based biometric system must be applicable and usable by making the number of electrodes and tasks that required as less as possible to reduce the Difficulties of EEG signal acquisition from the scalp of the user, as well as the complexity of the system, and the processing time should reduce by using feature extraction methods and classifiers with less computational complexity (Abo-Zahhad, Ahmed and Abbas 2015). In this paper the use of lowest number of tasks and channels (i.e., one channel per task) was tested to achieve high recognition rates without need to features fusion step, this keep the required computational complexity as low as possible. The discrete Fourier transform is applied, and a set Paper ID: ART20176484 DOI: 10.21275/ART20176484 46