Classification of driver drowsiness level using wireless EEG

Abstract

In this work, wireless Electroencephalogram (EEG) signals are used to classify the driver drowsiness levels (neutral, drowsy, high drowsy and sleep stage1) based on Discrete Wavelet Packet Transform (DWPT). Two statistical features (spectral centroid, and power spectral density) were extracted from four EEG frequency bands (delta, theta, alpha, and beta) using Fast Fourier Transform (FFT). These features are used to classify the driver drowsiness level using three classifiers namely, subtractive fuzzy clustering, probabilistic neural network, and K nearest neighbour. Results of this study indicates that the best average accuracy of 84.41% is achieved using subtractive fuzzy classifier based on power spectral density feature extracted by db4 wavelet function.