Research On Front Vehicle Detection And Distance Measurement Based On Machine Vision

The automobile industry has gradually integrated into people’s lives in recent hundred years.At the same time,the frequency of traffic accidents poses a great threat to people’s lives and property.The safety of automobiles has become a problem that needs to be solved urgently.With the development of science and technology,intelligent driving is becoming an increasingly important area,vision-based vehicle detection has become an important field,which aims to ensure the safety of passing vehicles by detecting front vehicle.This paper presents a method to detect the vehicle in front by using the combination of edge feature and shadow feature.The distance between vehicles will be obtained through the calibration of camera parameters and the transformation of different dimension coordinates.Finally,according to the safe distance of the vehicle to develop anti-collision warning strategy.Though by analyzing and comparing various algorithms on vehicle detection introduced by a large number of literatures of domestic and foreign,some approaches are proposed to improve the performance of vehicle detection algorithm.The main work of this dissertation shown as follows:(1)Image preprocessing.It takes a long time to detect a complete image,in order to improve the detection efficiency,select a valid area of 3/5 below the image to process.Grading the effective area by weighted average,which has closest gray level with the human eye,and then the median filter is used to eliminate the random noises generated during the data acquisition.(2)A method for identifying vehicles by using vehicle edge features is proposed.Which takes into account the different shape characteristics of the vehicle tail on the road,the contour of vehicle tail is extracted by Prewitt edge detection operator,the symmetry and length characteristics of the rear edge of the vehicle are then used to determine the area where the vehicle may be located.(3)Vehicle verification based on vehicle shadow characteristic.Two times using the Otsu threshold algorithm can effectively segment the vehicle shadow,then the candidate area of the vehicle is obtained by means of the morphological treatment of expansion and corrosion.The accurate verification of the vehicle is carried out by using the vehicle shadow shape characteristics and the positional relationship with the vehicle.(4)According to the principle of distance measurement of Monocular vision,the mutual transformation between different dimension coordinate system is established,then the internal and external parameters of the camera are calibrated.Through the transformation of coordinate relations and the parameters of camera calibration,based on the distance measurement formula deduced from the camera model,the distance measurement between the vehicle is completed.Finally,develop an early warning strategy.Through a large number of simulation experiments,it can effectively identify the vehicle in front.At the same time,the good data results are obtained by the distance measurement experiment,it shows that the method of distance mearsurement in this paper can meet the requirement of early warning,and has certain real-time and stability.