Research On Driving Behavior Analysis Based On Machine Vision

With the rapid growth of the number of vehicles,people's lives and property are threaten by more and more traffic accidents,some traffic accidents are caused by the driver's improper behavior.With the development of science and technology,it's possible that use various methods of detection to determine the driver's behavior.By detecting improper driving behavior and taking corresponding measures,the incidence of traffic accidents can be effectively reduced,so it has great significance for protecting people's lives and property.This paper studies the driving behavior analysis based on machine vision.In this paper,I carry out feature detection on the acquired lane line image and driver's face image,and fuse the preliminary conclusions obtained through the feature,finally obtain the driver's current driving behavior.status.The main work of this paper is as follows:1.Research on the detection of structured roads.The structured road detection is transformed into lane line detection,which is realized by filtering denoising,edge detection,Hough transform and so on.Then the lane line is tracked to enhance the robustness of the system.2.Research on the detection of unstructured roads.In this paper,the detection of unstructured roads is divided into three parts: texture main direction estimation,vanishing point detection and road boundarydetection.In the part of texture main direction estimation,for the principle of texture main direction estimation,this paper proposes a method of texture main direction estimation which is based on LDP texture feature calculation method.It can release the main direction estimation of image texture.In this paper,the method of combining depth learning with the traditional vanishing point detection method is proposed to improve the detection effect.In the part of road boundary detection,this process is divided into three parts: the first boundary detection,the vanishing point updating,the second boundary detects.The detection of road boundary line is realized by angle difference and color difference.3.Detection of driver facial features.In this paper,the features of face detection,eye detection,and lip detection are detected in images which are obtained by the on-board camera.Finally,their precise position is found in the image.4.Research on driving behavior analysis.In this paper,we study how to obtain driver's facial features and vehicle driving information by image.For travel information for vehicles,this paper proposes to judge the driving state of the vehicle by calculating the change rate of the distance between the center line of the lane and the center point of the image.In order to get a more accurate conclusion,this paper uses information fusion technology to get the final conclusion.In this paper,the detection of structured and unstructured roads,thedetection of driver's facial features and the analysis of driving behavior are realized.The results show that this method can accurately detect the current behavior of the driver and meet the requirements of the subject.