| Road traffic safety is related to the safety of everyone’s life and property,and the main culprit of most traffic accidents is fatigue driving.Therefore,how to effectively,timely and accurately detect the fatigue level of vehicle drivers and give corresponding fatigue intervention,so that the drivers can achieve the effect of conscious,has become the current research hotspot.However,the existing scheme is not widely used because of its low practicability,high requirements for drivers,real-time,accuracy and stability.In this paper,a fatigue detection and intervention scheme based on the side eye movement is proposed for the first time.A wearable fatigue detection device with glasses as the carrier is designed.Starting from the collection of the eye movement characteristics of the side eye,combined with the n-range eye location and eye state recognition method,the eye movement fatigue data is extracted accurately and in real time.According to the percentage of eye closed over criterion determines the fatigue level of users.The research contents of this paper include:(1)A new type of side eye movement Fatigue Inspection glasses is proposed.In view of the disadvantages of the current fixed fatigue detection scheme,such as low mobility,single use scene,low accuracy and low stability,this paper proposes a wearable fatigue detection glasses based on the side eye movement for the first time based on the principle of eye movement fatigue detection and user research and technical analysis.The camera collection position is placed at the side eye,and eye movement is collected through the side Image and identify the fatigue state of users.(2)This paper proposes an eye location algorithm.In order to solve the problems of large amount of computation,poor real-time performance and low accuracy caused by the two-stage face eye location,this paper combines the side eye images collected by the new fatigue detection glasses,uses the design advantages of the glasses to skip the face detection link and directly locate the human eye,and proposes an n-range eye location algorithm,which is verified by experiments It has high real-time and accuracy.In the process of side eye image processing,n-range algorithm uses n × n convolution kernel and sigmoid function to calculate the activation image,uses Otsu threshold segmentation method to segment the activation image,calculates the standard deviation projection of horizontal and vertical directions,and extracts the eye region from the projection image through the average threshold legal position.(3)In this paper,a scheme of eye state recognition is proposed.In the traditional positive fatigue detection scheme,eye area change is directly used to identify the eye state of others,which will lead to the change of eye area with the change of acquisition distance,thus greatly affecting the stability and accuracy of fatigue detection results.This paper proposes a method to recognize the eye state of others based on the change of the aspect ratio of the side eye,so as to enhance the mobility between the device and the driver,effectively resist the error caused by the relative displacement between the device and the person,and improve the stability of fatigue detection.(4)Organize experiments and training,assign the subjects to different working environments and different fatigue states in order to verify the effectiveness of the scheme in an all-round way,use p80 principle based on PERCLOS to conduct real-time fatigue analysis for the device users,and collect the user experience feedback of the subjects.The experimental results show that the existing problems of the current equipment have been well solved by the scheme proposed in this paper.The new fatigue detection glasses and n-range eye positioning algorithm can extract the fatigue features of eye movement accurately in real time and make fatigue judgment,and do not affect the normal operation of the user in the process of fatigue detection,which has high practical value and user experience. |
