Research On The Key Technologies Of Omni-directional Vehicle Vision System

Omni-directional vehicle vision system,also known as 360 panoramic system,is an important part of ADAS(Advanced Driver Assistance Systems).Omni-directional vehicle vision system is designed to present the object which around the vehicle body to drivers directly.By reducing the impact of visual blind spot on driving,the omni-directional vehicle vision system can guarantee the safety of driving which under the condition of limited vision,such as narrow road meeting,backing and so on.At present,there are still some problems existing among the mainstream vehicle panoramic systems such as blind spot,image dislocation and so on.Therefore,the purpose of this paper is to find a method to construct the system without blind spot and low dislocation by analyzing the construction process and exploring the construction difficulties of the system.In this paper,the construction process and main technical steps of the vehicle panoramic system are analyzed firstly.Based on the calibration of fish-eye camera and the contrast to ordinary perspective camera,a perspective correction scheme for fish-eye distortion image is constructed.By calibrating the actual installation position of fish-eye camera on the vehicle,the suggested installation angle of different installation height is given.Moreover,through analyzing the defects of common used method of bird-eye view transformation,a new transformation method is proposed this paper with the consideration of deflection effect to lay the foundation for subsequent image processing.Afterwards,the common used subsequent stitching method is applied in the field of panoramic image mosaic.And then the defects of subsequent stitching and the particularity of panoramic image stitching are analyzed.In order to overcome the difficulties in image mosaic such as dislocation and distortion,a new stitching method for panoramic image is proposed this paper.In addition,experiment is carried out to verify the proposed stitching method.Then,when it takes the practical application of the vehicle panoramic system into account,the phenomenon of panoramic image blur and poor visual effect which caused by uneven brightness is analyzed.In order to solve the problem,an image brightness equalization scheme is proposed and the brightness adjusting method is verified by the uneven light experiments.Finally,the experimental platform is designed and set up to simulate the actual vehicle size and camera installation.In addition,the experimental software is constructed by C#programming and the proposed image processing algorithms.By using the experimental platform,an experiment is carried out to build the omni-directional vehicle vision system in the real scenes.