Study On Fatigue Driving Behavior Detection Of Automotive Auxiliary Safety System

With the increase of car ownership in China,road traffic accidents occur frequently,which not only endanger people’s life safety,but also cause huge property losses.The fatigue driving and inattention of drivers gradually become one of the important reasons for the frequent occurrence of traffic accidents.In order to reduce the occurrence of such traffic accidents,it is of great practical significance to develop a set of reliable and effective fatigue detection methods.Judging the driving state of the driver by means of machine vision is the most promising detection method at present.This method has the advantages of non-contact,low cost and easy implementation.Among them,the driver’s facial features information can best reflect the driver’s fatigue state,this paper analyzes the face towards the relationship with the driver’s mental state,and then extract the reaction index of fatigue of eyes and mouth,three indicators and analysis of the relationship between the level of fatigue,fatigue calculation based on the fuzzy comprehensive evaluation model is established to determine the driver’s fatigue state.The main research contents are as follows:(1)This paper studies the theoretical basis of Haar feature based Adaboost algorithm,and uses Matlab to create a face classifier based on Adaboost algorithm to detect the driver’s face region in video images.For the detection of eyes and mouth area,this paper also adopts the Adaboost detection method based on Haar feature,and carries out the detection of eyes and mouth area on the basis of the obtained face area.(2)Face toward the driver is reflected when the vehicle is in the important process of fatigue driving,inattention,aiming at this problem,this paper builds the LVQ neural network,the design steps of face toward the recognition of the image into the neural network training and learning,finally recognize the driver’s face towards the video.When the number of images in which the driver’s face is not straight forward is calculated to exceed a certain number of frames,the driver may be in a state of fatigue or inconcentration.(3)The eyes and mouth images are preprocessed by binarization and morphological processing to highlight the features of the target image and reduce the interference of non-target regions to the image state recognition.The minimum rectangular aspect ratio of the eye area and the proportion of black pixels were used to judge the state of the eyes.The roundness method was used to judge the state of the mouth.The sample images were processed to get the threshold range of the eyes and mouth in different states.(4)After getting the state of the eyes and mouth in each frame of the video,calculate the driver’s blink frequency,PERCLOS value and yawning frequency index,analysis of the three indicators and the relationship between the level of fatigue,fatigue can be divided into three levels,using ahp to determine the weights of three indexes,fatigue calculation based on the fuzzy comprehensive evaluation model is established to determine whether the driver is in a state of fatigue.The above proposed method is tested to verify the accuracy and effectiveness of the proposed fatigue detection algorithm.