A new multi-level approach to EEG based human authentication using eye blinking☆
Introduction
Over the past 5 years, more than 100 publications have investigated the potential of the electrically recorded brain waves, known as Electro-Encephalo-Gram (EEG) signals, as biometric trait for human authentication purposes. EEG signals have gained a lot of attention in this field as they have some advantages over conventional biometric traits like finger-print and face recognition techniques. These traits are faced with the issue of low security. For instance, a finger-print system can be easily falsified using artificially generated finger-prints known as “gummy fingers” [20]. Moreover, face recognition technique can be easily spoofed using printed face models [10], [11]. However, EEG signals show more security since they are less likely to be artificially generated and fed to a sensor to spoof it. This also helps in addressing the liveness detection issue. Therefore, EEG biometric is expected to address some deficiencies of other biometric modalities or to complement them. EEG signals can be combined with other modality like finger-print for high security applications. Furthermore, when using EEG based recognition systems, it is impossible for an intruder to force a user to authenticate. In fact stress signals would be present in the measured brain waves, thus resulting in a denial of access.
The previous works in the field of EEG based biometric authentications have treated eye blinking Electro-Oculo-Gram (EOG) signals as artifacts and employed different algorithms for the removal of eye blinking and movement artifacts from EEG signals. However, an earlier study was conducted by the authors to examine the capability of eye blinking EOG signals to discriminate between individuals [2]. The eye blinking EOG signals were extracted from brain waves recorded using Neurosky Mindwave headset. This new modality for authentication has achieved high recognition rate up to 97.3% in identification mode and low Equal Error Rate (EER) of 3.7% in authentication (verification) mode over a database of 25 subjects. This encouraged the authors, in this letter, to propose a new multi-level EEG based biometric authentication system based on eye blinking EOG signals extracted from brain waves in order to improve the performance of EEG biometric authentication systems. Two approaches are adopted for the proposed multi-level system based on feature and score level fusion of EEG signals (during relaxation with eyes closed or visual stimulation) and eye blinking EOG signals. Density based and Canonical Correlation Analysis (CCA) strategies are applied for the score and feature level fusions, respectively. In the next two paragraphs, a brief background on EEG and eye blinking EOG signals is provided.
EEG is the electrical recording of brain activity, represented as voltage fluctuations resulting from ionic current flows within the neurons of the brain [21]. EEG can be recorded by electrodes placed on the scalp over the brain (non-invasive). The amplitude of the EEG signals ranges between 10–200 μV with a frequency falling in the range 0.5–45 Hz. EEG waveform is classified into five different frequency bands (alpha, beta, theta, delta, and gamma bands) [5]. Delta (δ) waves [0.5–4 Hz] are the slowest EEG waves, normally detected during the deep and unconscious sleep. Theta (θ) waves [4–8 Hz] are observed during some states of sleep and quiet focus. Alpha (α) band [8–14 Hz] originates during periods of relaxation with eyes closed but still awake. Beta (β) band [14–30 Hz] originates during normal consciousness and active concentration. Finally, Gamma (γ) waves [over 30 Hz] which are known to have stronger electrical signals in response to visual stimulation.
The eyeball acts as a dipole with a positive pole oriented anteriorly (cornea) and a negative pole oriented posteriorly (retina) [3]. When the eyeball rotates about its axis, it generates a large amplitude electric signal, which is detectable by electrodes near the eye known as electro-oculogram [8]. As shown in Fig. 1, when the eyeball rotates upwards, the positive pole (cornea) becomes closer to Fp1 electrode and produces a positive deflection. Similarly, when the eyeball rotates downwards, the positive pole (cornea) becomes far away from Fp1 (closer to the reference electrode) producing a negative deflection. This is similar to what happens when the eye blinks. When the eyelid closes, the cornea becomes closer to Fp1 and a positive deflection is produced. But, when the eyelid opens the cornea rotate away from Fp1 and a negative deflection is produced as shown in Fig. 2.
The outline of this letter is organized as follows. Section 2 presents a brief discussion on the related work. Section 3 describes the proposed multi-level EEG based biometric authentication using feature and score level fusion techniques. The experimental setup and the achieved results for different parameter settings are provided in Section 4. Finally, Section 5 summarizes the main findings and future directions.
Section snippets
Related work
The work done in the field of EEG biometrics can be grouped into three categories according to the type of EEG acquisition protocol used in the authentication task; EEG recordings while relaxations with eyes closed or open [6], [16], [28], [29], EEG recordings while being exposed to visual stimulation which are known as Visual Evoked Potentials (VEPs) [17], [24], and EEG recordings while performing mental tasks like imagination of moving fingers, imagined speech, imagination of rotating
Proposed multi-level EEG system using eye blinking
Any biometric authentication system consists of four basic modules; data acquisition, pre-processing, feature extraction, and classifier module [7]. However, for a multi-level system, a fusion module is added at raw data, feature, or decision levels. The two proposed approaches for the EEG multi-level system, as illustrated in Fig. 3, are discussed hereafter.
Experimental setup and results
In this section, the performance of the proposed multi-level biometric authentication system is evaluated. For evaluation, every time the experiment is run, 50 EEG frames (for relaxation or VEPs, where each frame is of one second duration) and 50 eye blinks are selected randomly for each subject out of 31 subjects. Then, features are extracted from the selected EEG frames (Section 3.3) and eye blinks (Section 3.2) to generate 50 feature vectors for each subject. After that, 40 out of 50 feature
Conclusion and future directions
This letter introduced a new multi-level approach for human authentication using brainwaves based on eye blinking EOG artifacts. In a previous work, eye blinking EOG signals have been shown to carry discriminant features that are capable for human recognition [2]. These features have been adopted here to improve the accuracy of EEG biometric authentication systems. The proposed multi-level system is based on fusing eye blinking with EEG at the feature and score levels. For the same number of
References (41)
- et al.
Dipole models of eye movements and blinks
Electroencephalogr. Clin. Neurophysiol.
(1991) - et al.
Biometric verification of a subject through eye movements
Comput. Biol. Med.
(2013) - et al.
Visual stimulation alters local 40-hz responses in humans: an eeg-study
Neurosci. Lett.
(1995) - et al.
A review of parametric modelling techniques for eeg analysis
Med. Eng. Phys.
(1996) - et al.
Information fusion in biometrics
Pattern Recogn. Lett.
(2003) - et al.
A theorem on the generalized canonical projective vectors
Pattern Recogn.
(2005) - et al.
A new method of feature fusion and its application in image recognition
Pattern Recogn.
(2005) - et al.
Pcg biometric identification system based on feature level fusion using canonical correlation analysis
Proceedings of the IEEE 27th Canadian Conference on Electrical and Computer Engineering (CCECE)
(2014) - et al.
A novel biometric approach for human identification and verification using eye blinking signal
IEEE Signal Process. Lett.
(2015) - et al.
Subject identification from electroencephalogram (eeg) signals during imagined speech
Proceedings of the Fourth IEEE International Conference on Biometrics: Theory Applications and Systems (BTAS)
(2010)
Brain waves for automatic biometric-based user recognition
IEEE Trans. Inf. Forensics Secur.,
Brain waves based user recognition using the eyes closed resting conditions protocol
Proceedings of the IEEE International Workshop on Information Forensics and Security (WIFS)
A survey of biometric recognition methods
Proceedings of the 46th International Symposium Electronics in Marine, 2004. Proceedings Elmar 2004
The eye blink electro-oculogram
Br. J. Ophthalmol.
Pattern Classification
Spoofing face recognition with 3d masks
IEEE Trans. Inf. Forensics Secur.
Three-dimensional and two-and-a-half-dimensional face recognition spoofing using three-dimensional printed models
IET Biom.
Modulation of induced gamma band responses in a perceptual learning task in the human eeg
J. Cogn. Neurosci.
Canonical correlation analysis: an overview with application to learning methods
Neural Comput.
Eeg spectral analysis of relaxation techniques
Appl. Psychophysiol. Biofeedback
Cited by (0)
- ☆
This paper has been recommended for acceptance by Maria De Marsico.