Subjects identification using EEG-recorded imagined speech

摘要: Abstract Due to the problems presented in current traditional/biometric security systems, interest use new have been increasing. This paper explores of brain signals EEG-based during imagined speech order it as a biometric measure for Subjects identification and thus create system. The main contribution this are two methods feature extraction, first improve signal-to-noise ratio Common Average Reference was applied. method based on Discrete Wavelet Transform, second statistical features directly from raw signal. proposed were tested dataset 27 who performed 33 repetitions 5 words Spanish. results show feasibility task with accurate Subject, regardless word used using commercial EEG system (EMOTIV EPOC). In addition, scope is displayed by decreasing training data, well number active sensors task. Using future improvements implementing low-cost device can be valuable