VIGILANCE ANALYSIS BASED ON EEG SIGNALS: SEEKING FOR SUITABLE FEATURES

Electroencephalography (EEG) is considered a reliable indicator of a person's vigilance level. In this paper, we use EEG recordings to discriminate three vigilance states of a person, namely alert, drowsy, and sleep, while driving a car in a simulation environment. EEG signals are recorded and divided into five-second long trials. From these EEG trials, we extract feature vectors containing a large set of features. Random forest is used to rank the plenty of features and select the most important ones for later classification. After dimension reduction, sample vectors are trained and classified by Support Vector Machine (SVM). The proposed framework explores different methods of EEG signal processing to discover the most suitable features for a real-time vigilance monitoring system. We investigate and compare three different kinds of features which are based on Continuous Wavelet Transform (CWT), Discrete Wavelet Transform (DWT), and Fractal Dimension (FD), respectively. On datasets acquired from 5 subjects, our result shows the CWT-based features reveal the highest classification accuracy (may reach over 96%). The DWT and FD-based features are less time-consuming in computation, and also reveal good result of classification accuracy (over 90%).