Drowsiness Detection based on EEG Signal analysis using EMD and trained Neural Network

Detection of drowsiness based on extraction of IMFs from EEG signal using EMD process and characterizing the features using trained Artificial Neural Network (ANN) is introduced in this paper. Our subjects are 8 volunteers who have not slept for last 24 hour due to travelling. EEG signal was recorded when the subject is sitting on a chair facing video camera and are obliged to see camera only. ANN is trained using a utility made in Matlab to mark the EEG data for drowsy state and awaked state and then extract IMFs of marked data using EMD to prepare feature inputs for Neural Network. Once the neural network is trained, IMFs of New subjects EEG Signals is given as input and ANN will give output in two different states i. e. drowsy or awake. The system is tested on 8 different subjects and it provided good results with more than 84.8 % of correct detection of drowsy states.