BibTeX entry

@PHDTHESIS{200702Daniel_Parnham,
  AUTHOR={Daniel Parnham},
  TITLE={An Infrastructure for Video-Augmented Environments},
  SCHOOL={University of York},
  MONTH=Feb,
  YEAR=2007,
  URL={http://www.bmva.org/theses/2007/2007-parnham.pdf},
}

Abstract

The development of a Video-Augmented Environment requires an integrated set of programming and image processing elements. Each of these elements is presented here as both an independent subject and part of a coherent infrastructure. The primary aim of this research was to realise an infrastructure that would provide a basis for the rapid development of Video-Augmented Environment applications. The OpenIllusionist framework is introduced as an embodiment of this infrastructure and provides a platform for further research and stable application development. This thesis presents agent-based design as an efficient technique for development. It then explains how a framework can support agents along with other necessary infrastructure elements. Fiducials are suggested as a means for both calibration and interaction. A new fiducial design is presented and shown to be more reliable than that used in a large proportion of augmented reality research. Automated and efficient calibration algorithms are introduced that are straightforward to incorporate into a framework. Shadow removal is discussed and then one particular technique is adapted for use in a finger detection algorithm providing an improvement over background subtraction methods. Finally a practical realisation of a Video-Augmented Environment is presented that was developed using OpenIllusionist. Overall, this thesis not only shows the feasibility of rapid VAE development, but also provides a practical framework, comprising a range of experimentally-verified technologies, to support efficient and robust applications.