Driver Fatigue Detection based on Eye State Analysis

Abstract

In this paper, we present an effective vision-based driver fatigue detection method. Firstly, the interframe difference approach binding color information is used to detect face. If exists, the face area is segmented from the image based on a mixed skin tone model. Then we simulate the process of crystallization to obtain the location of eyes within face area. Later, eye area, average height of the pupil and width to height ratio are used to analyze the eye's status. Finally, the driver fatigue is confirmed by analyzing the changes of eye's states. The experimental results show validity of our proposed method.