Study On Real-time Fatigue Driving Detection System Based On Eye Features

With the rapid development of transportation,the number of traffic accidents is rising year by year,which have seriously affected the people's life and property safety.Fatigue driving is one major cause of traffic accidents.In recent years,fatigue driving is more likely to occur because cars are designed to meet the requirements of comfort.So,it is necessary to find out a robust method to detect driver's fatigue level and give a warning.Nowadays,all countries in the world attach great importance to develop the driver's fatigue early warning device.Starting in the 1990 s,fatigue driving detection method based on various physical sensors became keystone of research on fatigue driving warning system.Then in 21 centuries,with the rapid development of computer science,the warning system have greatly improved in real-time performance,robustness and integration.But still,there are many problems remain exist: 1.Restricted by monitoring instrument,made it hard to perform realtime monitoring.2.Single detection indicators of detection methods had limitations.3.The detection device cost is too high,which is unfavorable for commercialization.In this paper,we propose a new method to detect driver's fatigue level.The main work of this paper is as follows:1)Object detection.By comparing the performance of different face detection methods,we choose the appropriate learning algorithm for real-time detection.And solve problems like occlusion,torsion and skin-color interference by introducing an improved Camshift.2)Fatigue features extraction.Set ROI regions of eye image according to the standard facial structure,and use techniques such as image enhancement and image segmentation to extract eye fatigue features.3)Decision making system.The PERCLOS characteristic value is calculated to judge the driving fatigue state,and divide driver's fatigue level into 3 levels based on eye blink duration.In this paper,our detection system is developed by Visual Studio 2010 and OpenCV,realtime experiment has proved that our system have good robustness and real-time performance.In addition,this article also carried out research on night monitoring for fatigue driving,and achieved good results.