EEG-Based Real-Time Drowsiness Detection Using Hilbert-Huang Transform

Human drowsiness detection is of vital importance in attention-demanding tasks. Electroencephalogram (EEG) signals have close relations with human vigilance level and thus are widely used in relevant research. In this paper, a novel approach for real-time drowsiness detection based on EEG signals is proposed. In comparison with previously proposed methods based on fast Fourier transform (FFT) or continuous wavelet transform (CWT), the proposed one makes use of the advantage of Hilber-Huang transform (HHT) in processing nonlinear and non-stationary signals, especially biological signals like EEG, to achieve more reliable time-frequency analysis results. The experiment shows that the HHT-based method actually renders more precise drowsiness detection results than the other two. Also, the proposed method uses single-channel EEG signals instead of multi-channel ones entailing cumbersome sensor systems. Such a characteristic facilitates the development of wearable drowsiness detection devices for daily use.