Research On Pre-vehicle Warning Technology Based On Radar And Monocular Vision

At present,as the number of cars increases.,road traffic accidents have become more frequent.In these traffic accidents,the rear-end collision of automobiles accounts for a large proportion,seriously threatening people's personal and property safety.Therefore,in order to effectively avoid the occurrence of such traffic accidents,developing a vehicle detection and warning system has become an urgent task.Common vehicle detection methods have two methods.one is based on machine vision and another is based on radar.Front vehicle detection based on Machine vision has good robustness,but it has poor real-time performance and is easily affected by ambient light and other environmental influences,which affects the accuracy of results.Front vehicle detection based on radar is not easily affected by the environment and has high real-time performance,but it is susceptible to noise and has a large atmospheric transmission loss.For traditional vehicle detection methods rely on a single sensor,there are problems such as missed detection and false detection.This paper proposes a front vehicle detection method based on a fusion of radar and machine vision.Firstly,I will use the CCD camera which mounts on the vehicle to obtain the video image of the vehicle in front,then send the video image to the PC,combining lane lines and vehicle shadows to initially obtain rough positioning of the front vehicle.Then determine the position of the front vehicle in the pixel space by detecting the left and right edge features of the vehicle and the gray value of the wheel.Secondly,use the radar to acquire the information of the obstacle in front,including the distance,speed,and acceleration of the effective target vehicle by filtering out the empty target,invalid target,and stationary target.Finally,fusion of radar information and visual information in time and space which is achieved based on the information fusion model.In this paper,after using the millimeter-wave radar and machine vision to detect the safety of the front vehicle,the information obtained by machine vision and the millimeter-wave radar information are merged.Based on information fusion,I will use the specialized Simulink module in the Matlab simulation software to establish the forewarning model,and build a simulation and warning monitoring simulation platform which based on the GUI image interface.Through a joint simulation experiment of Simulink and GUI,verified the warning al-Gorithm which based on target state prediction.Experimental results and data show that the method used in this paper can effectively reduce the false and false detection rate and can improve the accuracy of the detection results of the preceding vehicle,and has high real-time,robustness and accuracy.