Research On Mixed Reality Vehicle Driving Simulator Based On 3D Registration Technology

As a device that simulates driving behavior and driving experience,vehicle driving simulator has important research significance in the fields of traffic simulation,driving training,and vehicle design.A complete vehicle driving simulator consists of driving equipment,visual simulation system,control system and display equipment.Among them,the visual simulation system is the most critical,which determines the fidelity and real-time of the vehicle driving simulator.Mixed reality technology can mix and display virtual and real scene images.Therefore,the combination of vehicle driving simulator and mixed reality technology can solve the problem that vehicle driving simulator cannot take into account the immersive experience of virtual and real environments.But in the process of mixed reality,there are some problems,such as low accuracy of virtual and real object superposition,low accuracy of virtual and real image mixing and low display efficiency.These problems can be effectively dealt with by 3D registration technology.Based on the above background,this thesis designs and implements a mixed reality vehicle driving simulator based on 3D registration technology,relying on vehicle driving simulator and combining with 3D registration technology in mixed reality.Based on the research of vehicle driving simulator and the basic principle of mixed reality,this thesis presents the design scheme of mixed reality vehicle driving simulator system,and determines three stages of feature point extraction and matching,camera pose estimation and image registration in the 3D registration module of visual simulation system.In this thesis,the three stages are studied respectively: firstly,feature point extraction and matching algorithms,SIFT,SURF and ORB,are studied.Through experimental comparison and result analysis,ORB is used as the feature point extraction and matching algorithm of the system,and RANSAC is introduced to filter out the wrong matching pairs.Then,this thesis studies the camera calibration and imaging principle,uses the camera pose estimation method based on image feature points to solve the camera external parameters,and then combines the feature point information to realize the superposition of virtual markers on the real image.Finally,this thesis studies the image registration in mixed reality.Using the image registration method based on feature points,the geometric transformation of the registered image is carried out to realize the accurate mixing of virtual and real images,and the brightness adjustment in mixed reality is studied.HSV is used to adjust the brightness to keep the brightness display effect of mixed reality close to the real scene.At the end of this thesis,the algorithm of each part is tested and analyzed,and the smooth and stable display of the mixed reality vehicle driving simulator system based on 3D registration technology is realized by integrating each stage with the vehicle driving simulator.Through the experimental results and data,the correctness and efficiency of the mixed reality vehicle driving simulator system and related algorithms are verified.