Wavelet based Analysis of EEG Signal for Detecting various Conditions of Driver

In this paper, driver's drowsiness related conditions are detected using Discrete Wavelet Transform (DWT) from the ElectroEncephaloGram (EEG) signal (0.1 - 100 Hz) collected from the brainsense headset. Data is collected by mounting the headset on driver's head on various sessions of a day i.e., during morning, afternoon, night and when a high beam torch is focused into eyes of a driver. The obtained data sets record length is of 23.6 seconds and sampled at 173 samples/second. From the EEG signal; alpha, delta and theta waves ranging between 0.1 to 12Hz is extracted using various levels of DWT decomposition technique. This helps in detecting various conditions such as unconsciousness, drowsiness, dreaming state. An algorithm based on DWT and Support Vector Machine (SVM) classifier is developed to detect the various conditions, purely based on eye state analysis of a driver. This is then tested with 20 eyes opened and 20 eyes closed datasets where each data set record length are of 4097 samples. A 95% of accuracy is obtained from the developed algorithm.