Part-Based Templet Matching in the Detection of Fatigue Driving

We describe a face detection system based on multiscale part models for drivers´ fatigue detection. Our system is able to represent human face using two eye areas and one mouth area. By using mulriple simple template instead of single template with complex details, new model observably improved the computational efficiency and veracity for face detection. This system also considers the specialty of fatigue detection and put out a new method for face detection for it. While color recognition is the most effective and simply method for face detection in conventional face detection. The specific application and condition decide the advantage of the model in the detection fatigue of drivers. To reflect the characteristic of the features of fatigued riving discriminant background, this paper uses an image with a complex background to simulate.