Picture-based Localisation For Pervasive Gaming
Virtuelle und Erweiterte Realität, 11. Workshop der GI-Fachgruppe VR/AR, (2014)Abstract
Localisation, i.e. determining the position of the user(s) or devices, constitutes the key requirement for almost all types of mobile pervasive games. However, current marker-based localisation systems, e.g.\ based on QR Markers, present drawbacks that limit game deployment, scalability and maintainability. In this paper, we propose an alternative to solve these issues and introduce the first steps towards its full realisation. Our approach relies on markerless picture matching using a natural feature detection algorithm. Players reproduce a camera shot of a real-world site in order to confirm their presence, and progress further in the game. One of the game critical requirements is to provide accurate recognition while preserving application responsiveness with a large range of mobile devices and camera resolutions. We developed a proof-of-concept system and determined the best picture resolutions and feature numbers necessary to preserve both accurracy and responsiveness on diverse mobile devices. Our first results demonstrate the feasibility to achieve precise recognition within real-time constraints. We believe such localisation system have the potential to considerably facilitate pervasive game authoring while promoting new type of game mechanics.