Research On Driver Fatigue Detection Algorithm Based On Facial Feature Points

With the development of society and economy,the number of cars is more and more,which provides people with all kinds of conveniences and brings great pressure to the safety of travel.From the statistical survey of traffic accidents,it is found that traffic accidents caused by driver fatigue are increasing year by year.Therefore,it is very important to detect the fatigue state of the driver in real-time.It is urgent to develop a driver fatigue monitoring system,which can detect and warn early and reduce traffic accidents.There are a lot of research on fatigue detection institutions in the world at present,there are many detection systems are being developed,but most practicality is not high,the detection accuracy and efficiency is not high,generally used to need a long time development.In this paper,the methods of fatigue detection at home and abroad are compared,and the advantages and disadvantages of each method are analyzed.A new method based on visual image is presented.The method to analysis the facial image by using image processing technology,the use of facial feature extraction can reflect the fatigue characteristics of the eyes and mouth information,combined with the improved PRECLOS and the fatigue state detection and warning system of fuzzy neural network to the driver,and achieved very good results in terms of recognition rate and speed.The main contents of this paper are as follows:(1)A detailed analysis of the current domestic and international research on driver fatigue detection methods,through the detection accuracy and practicality of contrast,the decision to use visual image method for face detection to achieve fatigue judgment.(2)The Haar cascade algorithm and the HOG cascade algorithm are studied,and the contrast experiment is done,and the face detection algorithm is improved.The experimental results show that the improved face detection algorithm can achieve fast and accurate face detection.(3)A feature extraction algorithm based on facial feature points is studied.The regression method will learn a regression tree directly from the image appearance feature to the target output shape.Each regression function is learned by the gradient algorithm,and the regression tree is used to fit the residual error.Therefore,the regression tree can be used to locate the feature points from sparse images.(4)To extract the information of face feature points,through the improved PRECLOS algorithm to calculate the blink of an eye,a blink of an eye,and a yawn yawn four fatigue parameters,and then input into fuzzy neural network system,after fuzzy processing,the fuzzy rules to judge the fatigue state of driver output.