Font Size: a A A

Design And Implementation Of Mobile Augmented Reality System Based On Monocular 3D Object SLAM

Posted on:2022-08-25Degree:MasterType:Thesis
Country:ChinaCandidate:R TaoFull Text:PDF
GTID:2518306341451984Subject:Computer technology
Abstract/Summary:PDF Full Text Request
With the development of science and technology,people have put forward higher requirements for mobile augmented reality technology.However,traditional augmented reality simply superimposes virtual objects on top of the real world and cannot truly integrate virtual objects with the real world in the presence of occlusion.Aiming at the problems of traditional mobile augmented reality,this paper designs and implements a mobile augmented reality system based on 3D object SLAM,which uses a lightweight semantic SLAM algorithm to map the surrounding environment and represents real-world objects in the form of cubes.The system constructs a map with semantic information to solve the occlusion problem of AR.This paper fully considers the uncertain initialization pose of smart mobile phones and the problem that the monocular camera equipped with Android devices cannot restore the absolute scale.An initialization algorithm based on the fusion of inertial sensors and 3D objects is proposed to combine visual data with inertial sensors is proposed,which align the visual data with inertial measurement unit data to restore the initial pose and scale of the 3D objects,so that the accuracy of the system is improved.The experiments of this algorithm on public data sets and the real-world show that this algorithm can achieve better results and can ensure sufficient computing speed.Based on the improved 3D object SLAM algorithm,this paper designs a mobile augmented reality system running on Android devices.According to the system architecture,the system is divided into 2D detection module,pose initialization module,3D cube generation module,SLAM fusion module and AR interaction rendering module,each module contains several sub-modules.The function test and performance test of the system show that the system can construct a map of the real world environment in real time,estimate the pose of the camera,and realize the occluded augmented reality effect,which has sufficient effectiveness and stability.
Keywords/Search Tags:mobile augmented reality, SLAM, pose estimation, IMU
PDF Full Text Request
Related items