Research On Driver Fatigue Detection Technology Based On Eye State

With the rapid increase of the number of cars, the vicious traffic accidents increased significantly. The main reason is that the driver fatigue or distraction of their mind. Driver fatigue detection systems can estimates the moment that the driver fatigue in advance. Therefore, the driver fatigue detection systems have become the focus of research and development all over the world.The real-time detection for the driver fatigue is an effective way to improve the security of road. And it is an important factor in driver fatigue detection system to find the facial image、train classifier and judge the eyes’statement. The image preprocessing is been token to remove noise, improve the quality of the images and keep image characteristics.A new sub-level light compensation scheme is proposed because of the complexity of driving environment. The median filtering is used to remove image noises. An improved algorithm syncretized the skin color information is put forward to divide the picture. And then, use the canny algorithm to remove much background information. They make the machine classification training a lot easier. The Adaboost algorithm leads to many samples becoming wrong and reducing the accuracy rate. So a new improved Adaboost algorithm is proposed. After the eye detection, the eyes state is judged by the ratio between vertical distance and horizontal distance of the eyes.Using the PERCLOS algorithm to judge the level of fatigue. Finally, conduct experimental verification to prove the improved algorithm can increase the speed of the recognition and the accuracy rate of system.This study will play a positive role in the development of real-time detection for the driver fatigue.