Research On The Around View Monitor System Of Multiple Units Train

With the increase of vehicle ownership in China,traffic accidents happen frequently.The main reason for traffic accidents is that the drivers are not familiar with the surrounding environment.The Around View Monitor System can display the surrounding environment of the vehicle to the driver to assist in driving in real time,so as to reduce the accident rate.At present,the Around View Monitor System on the market is aimed at passenger cars or single-unit buses,which is not suitable for multi-group vehicles included trailers,articulated buses,and trains.However,multi-group vehicles are huge,and only rely on the rearview mirror to have a large field of vision blind area.It is very difficult to ensure driving safety.The use of multi-group vehicles is to transport large quantities or passengers.Once a traffic accident happens,it often causes greater losses.Based on this,this paper studies and develops Around View Monitor System for multi-group vehicle.This system obtains images through fisheye cameras and processes image processing algorithms to display the bird's-eye view image around the vehicle in real time.The main research work of this thesis is as follows:1.The generation of single car panoramic view image.Through fish eye image acquisition,distortion correction,perspective transformation and image mosaic,a look-up table representing the mapping relationship between the distorted image and the panoramic bird view is generated.In the course of fish eye image acquisition,the wide field of fisheye camera is selected.In the process of distortion correction,the polynomial projection model is deeply studied.In the process of perspective transformation,the single stress matrix is derived,and a method of automatic selection of reference points with MR operator and SUSAN operator is proposed.In the process of image mosaic,completed the matching of feature points based on the checkerboard.2.The splicing of panoramic view image for multi-group vehicles.This paper import the image information into GPU by the way of texture mapping.Through the establishment of multi compartment panoramic image mosaic model and the analytic CAN angle signal,the real-time dynamic splicing of multi view vehicle panoramic view image is realized.3.The transplant of the algorithm and the test.The algorithm is transplanted on the development board with IMX6 as the core board,and the installation test is carried out on a vehicle with three units.The result shows that the panoramic view image splicing has a good effect.The research show that the multi-group variable-angle vehicle panoramic view system based on angle sensor developed in this paper solves the problem of real-time panoramic splicing systems and methods with a large number of cameras and relative position change of the camera.This system can assist driver driving and has a good application prospect and reference value.