Abstract
A drowsy driver detection system based on a new method for head posture estimation is proposed. In the first part, we introduced six possible models of head positions that can be detected by our algorithm which is explained in the second part. Indeed, there are three key stages characterizing our method: First of all, we proceed with driver's face detection by Viola and Jones algorithm. Then, we extract the image reference and the non image reference coordinates from the face bounding's box. Finally, based on measuring both the head inclination's angle and distances between the extracted coordinates, we classify the head state (normal or inclined). Test results demonstrate that the proposed system can efficiently measure the aforementioned parameters and detect the head state as a sign of driver's drowsiness.. . .