Vehicle-mounted Driver Fatigue Monitoring System Based On Android Platform

With the increasingly perfecting of the road traffic infrastructure, the automobile volume and the number of automotive vehicle driver are increasing rapidly, fatigue driving is also becoming more and more serious, and it has gradually become one of the main factors leading to traffic accidents. Research shows that if fatigue state of the driver can be monitored in real time, and the corresponding warning reminder can be given timely in the early fatigue driving, a lot of accidents can be effectively avoid.In this paper,we have studied a vehicle-mounted driver fatigue monitoring system which based on android platform. This fatigue monitoring system takes android smartphone as hardware platform, and captures the driver’s images with the front camera of the android smartphone, and then measure the state of the driver fatigue with the PERCLOS indicators by analyzing driver’s eye closure state. The main work of this paper is as follows.(1) Face detection and tracking. Constructed the face detection system based on Adaboost face detection algorithm and the face tracking system based on Kalman filtering. Collected a large number of vehicle environment images in the vehicle environment which was used as face negative samples.(2) Eye location. Firstly, on the basic of face detection, narrowed the detection range of the human eye location further according to the geometric distribution of human facial organs. Then, constructed the eye location system based on Adaboost algorithm. Finally, separated the eyebrows from the eyes by windowed gray-level integrated projection algorithm, thus getting the precise location of the human eye parts.(3) Fatigue recognition. Using PERCLOS method for detecting driver fatigue and two methods of eye state recognition based on eye area and upper eyelid curvature were mainly studied. In the process of eye state recognition based on eye area, a dynamic adjustment method for binary segmentation was proposed.(4) Implementation of vehicle-mounted driver fatigue monitoring system based on android platform. The system is realized based on OpenCV, and the test results show that the average time of a complete monitoring process is 85.15 ms. Therefore, the system can basically meet the requirements of real-time fatigue monitoring.