Research On Road Sight Distance Detection Technology Based On Monocular Vision

With the rapid development of road traffic,traffic safety problems have become increasingly prominent,a large number of studies have shown that highway sight distance is insufficient,and the driver's access to dangerous road information lags behind,which is one of the major causes of frequent occurrence of road traffic accidents.Therefore,to ensure the driver in the process of road transportation to get enough sight distance is the primary task of ensuring the safety of road traffic.Therefore,how to detect road distance timely and effectively,improve highway safety design,to remind the drivers reasonable speed,avoid the occurrence of traffic accidents,improve the operation of the highway safety and security will be an important way to improve road traffic safety problems.In order to obtain good highway sight distance detection results,this paper studies the image processing and monocular vision distance measurement,and proposes a highway sight distance detection algorithm based on the coordinate transformation of lane line feature points.The main research contents are as follows:(1)First of all,the threshold segmentation based on the traditional Otsu method will cause a certain viscosity problem between Lane area and road area.The lane gray distribution curve through repeated testing to determine the range of gray lane based on the lane gray range is set to two threshold double threshold segmentation.(2)Secondly,aiming at the interference of edge information in the real road environment and the lane boundary fuzzy problem caused by uneven illumination and lane fouling,this paper uses Sobel vertical gradient convolution template to enhance the edge of the lane,the definition of lane edge and the effect of lane extraction is improved,then aiming at the problem of line interference after the initial extraction of lane,the lane scanning method based on lane width is adopted for the two extraction.(3)Finally,according to the relevant knowledge of sight distance,the problem of the detection of sight distance is transformed into the problem of the detection of the lane length between the lane vanishing point and the vehicle.The transformation relation between coordinate systems in camera imaging is analyzed and calibration to obtain the camera intrinsic parameters using Zhang's method.Through the geometric projection and coordinate transformation relationship,the sight distance detection model based on coordinate transformation between lane characteristic points is deduced.The static real vehicle test is carried out.(4)Research shows: the vision distance detection algorithm based on monocular vision in free flow condition suggested in this paper can be used to detect the highway sight distance of straight lines and curves,the average error of highway sight distance detection is 5.2%.The research results can be applied to the safety evaluation of the operation Highway and the safety reminding to the driver during the driving process.