Design And Implementation Of Panoramic Image Generation Algorithm Based On Monocular Vision For Driving Assistance

With the increasing car ownership and the increasingly congested road environment,the convenience and safety of cars have faced more difficult problems and tests.There are some problems with incomplete view coverage or complex and expensive installation in the common reversing image system and panoramic image system.Under this background,this paper studies some key issues of panoramic image construction based on monocular vision,and completes the design and implementation of a method of building panoramic images of vehicles using the car rear single video data.The main work and innovation of this paper are as follows:(1)Aiming at the requirement of vehicle panoramic image stitching based on monocular vision,a method that is used for panoramic image generation based on monocular vision for driving assistance is designed,which has the functions of system calibration,data preservation,feature extraction and template matching.The verification platform of the method is implemented on the Windows platform.After a variety of pavement video tests,the results show that the method can effectively generate panoramic images of vehicles on the road with rich texture or traffic marking.(2)Two commonly used embedded development hardware platforms based on the Linux and Android operation systems are selected,which enhances the effectiveness of the above method and completes the corresponding program design and function development.After the actual tests on the car,the results show that the program which runs on the Android operation system is more effective and real-time.This paper implements a method of panoramic image generation based on monocular vision,which initially possesses the practicality on the Android operation system.