Driver Fatigue Detecting Research Based On Infrared Images

Driver fatigue has become one of the main factors in the causes of traffic accidents, the method using machine vision to detect driver fatigue have greater advantages than other monitoring method in real time,accuracy,no- contact,applicability and economic area and become a hot research.Based on the research of other people's work, this paper think night is the period driver often feel tired .So we get the positive driver images using a special infrared light and develop an effective driver fatigue detection method in accordance with the characteristics of infrared images. The whole method is divided into four processes: Face detection, Tracking of the human face, Eye location and Driver fatigue recognition. The study of this paper is as follows:(1) Face Detection is the preliminary work in the driver fatigue detection. The traditional methods are mostly based on skin color segmentation algorithm and the gray-scale projection method. This two methods have high requirements in illumination, and can not exactly locate human face in uneven illumination, lack of light or frequent changes of poses. This paper starts from rough detect, uses iterative algorithm to threshold image and then propose a new algorithm using vertical projection and regional connectivity to accurately locate faces based on advantages of a simple background and higher brightness of face region in infrared images. This method not only can accurately detect the positioning face, but also less affected by the light, not sensitive to changes in attitude. It solves the problem skin color segmentation and gray projection can not exactly position because of the affection of illumination.(2) Eye location is a key step in the driver fatigue detection. Traditional Hough transform and template match requires face image with clear verge information. It also need take great amount of computation and do not well in real-time aspect. Based on the detailed analysis of the characteristics of infrared images, this article improves the Harris corner detection algorithm, and apply it to the position of the pupil. This method avoids a complex computing, locating accurately and haves great algorithm speed, fully meet the real-time requirements.(3)According to PERCLOS principle, this paper using the method of calculating eye areas to monitor driver fatigue in real time. For driving at night is the high incidence of fatigue period, this method takes infrared light as the light resource and resolve the instability of dim light in night. All the algorithm this article taken is simple and effective, low complexity, not only be not sensitive to the impact of illumination, but also has better tolerance and robustness.