Research On The Vision-Based Driver Fatigue Real-Time Detection

Research on the judgment and avoidance of drivers' drowsiness plays a very important role in reducing traffic accidents. The visible expression of driver's fatigue is the drowsy state of eyes. So, it is essential to judge the fatigue state of driver's eyes rapidly.The thesis describes a real-time system based on computer vision technology for monitoring driver's facial features including the positions of face and eyes. The system starts with acquiring a piece of consecutive video images of the driver's face using a CCD WebCam. Then the driver's eyes in every frame are located and the eye status is detected by the fast and compact algorithm. Then the driver's degree of fatigue is determined without affecting the driver and prompt warnings will be generated when necessary.This paper mainly covers the following works:Methods of face detection, eye location, eye states recognition and some techniques of digital image treatment are studied. Based on the current algorithms, the algorithm adapted to the system is presented.In this paper, we firstly introduce an object detection scheme based on a boosted cascade of Haar-like features which is capable of processing images extremely rapidly and achieving high detection rates. When the algorithm is used to detect face, it yields a satisfying result. A new method using template matching is presented for the eye location after face detection. The experiments have proved that it is very effective.And then we apply the gray-level projection and Hough Transformation method to predict the eye states. Finally, according to PERCLOS criterion, we can judge whether the driver is drowsy or not.Many tests show that the system can detect eye states. So it can detect the degree of fatigue efficaciously. From the running effect of this system, the efficiency of this system is very high in the laboratory environment.