RGBD SLAM Based 3D Object Reconstruction and Tracking: Using Google ARCore

Abstract

The significant growth in the popularity of augmented reality and virtual reality in the fields of technology and entertainment has created new challenges and opportunities. Development platforms like Google’s ARCore and Apple’s ARKit have enabled increased growth in augmented as well as virtual reality application development. Especially in augmented reality applications, three-dimensional reconstruction and environmental observation have become an important area. The aim of this work is to study the features of ARCore and to analyze whether it is possible to carry out a three-dimensional reconstruction of an object or scene using a standard smartphone camera. The work is divided into five sections, and at the beginning of the work, the reasons for this research and previous work and methods related to this topic are briefly reviewed. The second section discusses the features and operating principles of ARCore. The third section discusses the important methods for machine vision and especially augmented reality, on which ARCore operates. The section explains how to find out the three-dimensional structure of a scenario using motion, a depth sensor, or multiple cameras. The methods are briefly described in terms of their operating principles and uses. At the end of the work, two separate applications that utilize ARCore in three-dimensional reconstruction and the results produced by these applications are presented. In the testing phase, the two applications will be tested on a smartphone, and at the end, these results will be compared with each other, as well as possible future uses and features of ARCore will be considered. Thus, it is possible to implement a three-dimensional reconstruction with the help of data collected by ARCore, but ARCore itself does not directly support the reconstruction. Separate applications are needed to perform the texture meshing and visualization.