Fast Scene Recognition and Camera Relocalisation for Wide Area Augmented Reality Systems

Tao Guan,Liya Duan,Yongjian Chen,Junqing Yu

doi:10.3390/s100606017

Tao Guan, Liya Duan + Show 2 more

Open Access

https://doi.org/10.3390/s100606017

Copy DOI

Abstract

This paper focuses on online scene learning and fast camera relocalisation which are two key problems currently limiting the performance of wide area augmented reality systems. Firstly, we propose to use adaptive random trees to deal with the online scene learning problem. The algorithm can provide more accurate recognition rates than traditional methods, especially with large scale workspaces. Secondly, we use the enhanced PROSAC algorithm to obtain a fast camera relocalisation method. Compared with traditional algorithms, our method can significantly reduce the computation complexity, which facilitates to a large degree the process of online camera relocalisation. Finally, we implement our algorithms in a multithreaded manner by using a parallel-computing scheme. Camera tracking, scene mapping, scene learning and relocalisation are separated into four threads by using multi-CPU hardware architecture. While providing real-time tracking performance, the resulting system also possesses the ability to track multiple maps simultaneously. Some experiments have been conducted to demonstrate the validity of our methods.

Highlights

The objective of augmented reality (AR) is to add virtual objects to real video sequences, allowing computer-generated imagery to be overlaid on the video in such a manner as to appear part of the viewed 3D scene [1,2]
To obtain a system with real-time tracking performance, we implement our algorithms in a multithreaded manner by using a parallel-computing scheme
We demonstrate that the tasks of scene learning and camera relocalisation can be separated as individual threads to further improve tracking performance

Summary

Introduction

The objective of augmented reality (AR) is to add virtual objects to real video sequences, allowing computer-generated imagery to be overlaid on the video in such a manner as to appear part of the viewed 3D scene [1,2]. Registration methods for wide area unprepared environments have attracted much attention [3,4,5,6]. These methods have several advantages compared with registration methods which depend on prior knowledge of the user’s environment. Tracking is not limited to the prepared scenes, users can walk anywhere they want and superimpose virtual objects dynamically, according to the requirements of the AR applications

Methods

Findings

Discussion

Conclusion