Research On Recognition Method Of Driving Fatigue Characteristics Based On Physiological Parameters

Fatigue driving has been one of the important causes of traffic accidents,especially hindering the safety of life and property of drivers.Nowadays,the technology mainly focuses on the identification,detection and prevention warning of fatigue driving.So this technology is developing.At present,there are three kinds of methods to identify fatigue state:the test method based on heartbeat pulse,but it is not often used because it needs to contact the driver,because this method affects the driver’s driving;the recognition method based on facial information,although it does not affect the driver’s driving,but this method is too single,the recognition rate is too low,and it is not accurate;According to the operation behavior and vehicle operation condition,the inspection method is real-time,low cost,simple and easy to implement,but it is easy to be disturbed,unreliable,and easy to cause false identification and missing of fatigue condition.For the problem that the detection technology of a single signal source is not stable and the external environment has strong interference,based on the previous research,combined with the existing data and machine learning,processing picture information,physiological signals and other fields,this paper proposes detection methods based on the facial and heartbeat pulse information,which are physiological information,to extract the eye features and pulse in the face The time domain,frequency domain and other information parameters of the pulse signal are combined with multiple features to construct the vector feature space,and the PCA and CVM based methods are used to carry out the final fatigue state recognizer.Finally,the purpose of human body recognition is achieved.The general contents of this paper are as follows:Using the method of basic color segmentation and gray-scale integral projection to achieve the solution of the characteristics of human eye state,based on the skin color segmentation to recognize the face contour in the picture,then in the recognition of the face,using gray-scale integral projection to distinguish the human eye,then calculate the characteristic parameters of PERCLOS(percentage of eyelid closure over the pupil over time),opening and closing degree and blink frequency to recognize the state of other people’s eyes.The method based on wavelet transform is used to calculate multiple parameters such as pulse signal and frequency domain.Based on the structure of DB6 wavelet,it can reduce the noise and filter the information of the pulse signal,and then use the adaptive threshold method to locate the special points of the filtered peak of the pulse signal.After identifying the information,it can extract the AVNM,SDNN,HF,LG and other special parameters of the pulse signal in time-frequency and frequency domain.