Research And Implementation Of Drivers’Fatigue Detection System Based On AAM

The safety driving assistant system based on computer vision has a broad application prospect in improving vehicle safety and reducing traffic accidents. Drivers’ fatigue detection start with avoiding fatigue driving, and effectively detect the degree of drivers, then gave warning and thus to reduce traffic accidents and make driving safe, it has an important practical significance.This thesis researched on divers’fatigue detection technology based on computer vision, after systemically studied and analyzed the present situation of the fatigue driving both at home and abroad, selected the appropriate facial features as the basis for evaluating fatigue driving, designed a drivers’fatigue detection system based on Active Appearance Model (AAM). In this paper, after the skin-color filtering, we used the Adaboost algorithm to detect face and eyes, and then used the AAM to locate the facial feature points, after that, LK-optical flow algorithm was used to track those facial feature points. In the tracking process, we used AAM in adjusting the tracking results in order to eliminate the tracking deviation. In thefeature extraction stage, this paper selected the eyes and mouth feature extraction to determine the edges of eyes, then got the height difference of the upper and lower eyelid, determined the eyes’ state. Finally, we calculated the indicators such as PERCLOS value, at the same time, according to the feature points stand for mouth in AAM, recorded the mumber of yawning. Finally, we judged the fatigue happened or not through these indicators.Finally, we realized a prototype system, and took test using typical video. The result indicates that the fatigue detection system with the method combined by AAM and Adaboost introduced by this thesis can precisely reflect different situations of the drivers in every process.