An Eye State Recognition Method for Drowsiness Detection

The issue of public driving safety has become more and more important. During long-distance driving, the driver's consciousness will decline and the probability of traffic accident will increase. In order to avoid this situation, finding some approaches to realize whether the driver is drowsy or not is very critical. Because the eye state is the key point in driver drowsiness detection, we propose a method to recognize the eye state. At first we use the Haar-like features and Adaboost classifiers to find the face location. Then we use the SVM classifier to find the eyes locations. In third step we calculate the LBP features for the image of the left eye. Finally we put the features into SVM classifier to recognize the eye state. The experimental results show that the proposed method is effective for eye state recognition and is therefore helpful for driver drowsiness detection.