Research On Fatigue Driving Detection Method Based On Facial Features

With the development of society,people’s living standards have been improved,and the car ownership has been growing,but the number of automobile traffic accidents is also very large.Relevant data shows that a large part of accidents are caused by drivers’ fatigue driving.When the driver is in the state of fatigue,it is not very good to deal with the road emergencies,especially when the driver enters the sleep state in the driving process,which is likely to cause irreparable consequences for the safety of life and property.Therefore,it is of great significance to carry out the research on fatigue driving detection system and detect and timely warn the driver’s state.Due to the fatigue driving detection method based on the analysis of driver’s facial features,compared with the detection method based on vehicle condition and driver’s physiological condition,the device has no direct contact with the driver,and the driver’s comfort is better without affecting the detection effect.Therefore,this paper focuses on the fatigue driving detection method based on the analysis of facial features,aiming at the fatigue state The relevant eye features and mouth features were studied,and the eye closure time,blink frequency,yawn times and other parameters were used as fatigue discrimination indicators to complete fatigue driving detection,the specific contents are as follows:(1)In order to deal with the complex environment in the car and improve the image quality,this paper studies some image preprocessing methods,such as grayscale,image filtering,histogram equalization and so on;(2)In order to ensure the accuracy of eye and mouth region location,this paper first carries out face location,and then carries out eye and mouth location in the acquired face region,instead of directly carrying out eye and mouth location on the 320*240 size input image,reducing the detection area.In face detection,the face detection method based on Haar like feature is selected,which has fast processing speed and low resource demand.At the same time,in order to further improve the processing speed of video stream,combined with the characteristics of relatively fixed driver activity area in the car,the tracking part selects the fdst algorithm based on correlation filtering,which combines face tracking with face detection,and improves the short-term face occlusion and head The detection failure caused by partial deflection can reduce the calculation of the system,ensure the detection speed of the system,and improve the accuracy and speed of face detection;(3)In this paper,the original method of "three courtyards and five eyes" is improved,which greatly reduces the scope of accurate positioning of eyes and mouth.In the case of slight head deflection,this method can still accurately complete the coarse positioning of mouth and eye.In the target area of coarse positioning,according to the projection distribution characteristics of binary image,the accurate positioning of eyes and mouth is completed.The binary image of mouth area is also completed For example,this paper proposes a method based on the pixel distribution features to complete the two classification of the normal closed state and yawning state of the mouth;for the binary image of the eye area,this paper proposes the proportion of the pixel in the eye area as the discrimination index,which can accurately distinguish the eye state according to the experimental results;(4)A fatigue driving detection system based on multi fatigue discriminant index is designed,which combines the characteristics of eye closure time,blink frequency,yawning frequency,and so on.When one of the abnormal indicators is called alarm,it ensures the robustness of the system and ensures the safe driving of the driver.