Lewis Baker, PhD student, Department of Computer Science, University of Otago
Localization and tracking of stationary users for Augmented Reality
Camera localization and tracking are two key components of many Augmented Reality (AR) systems. Localization methods determine the pose of a camera within some known coordinate system or model, and tracking is the process of updating the pose in real-time as the camera moves. In many mobile AR applications, the user is standing or sitting stationary while rotating a smartphone around their body. This kind of motion causes problems for many tracking systems due to the small translational movement, especially in large environments. In this presentation, I will discuss the limitations of state of the art tracking systems, and introduce a new tracking method based on a spherical motion constraint. I will also provide an overview of the current approaches to localization, and present our experiments in applying a traditional model-based localization method to noisy structure-from-motion models.
Lewis Baker is a PhD student in the Visual Computing group. He joined in 2015 and worked on projects like a vision-based Power Line Detection and an extension of the ARSandBox. In 2017, he started his PhD working on creating an Augmented Reality Sport Spectator system. Spectators may use mobile devices or wear head-mounted displays to see live visualisations, statistics, and commentaries. This project has many challenges such as localisation, effective visualisations, and registration (aligning virtual objects with the real world), which become much more difficult in large uncontrolled environments
Last modified:
This page is maintained by the seminar list administrator.