Computer Vision and Virtual Reality in the UK
by Bob Fisher
Computer Vision is a well established research field in the UK, with perhaps the second largest and most productive research community outside of the United States. Recently, Virtual Reality has also become a significant research and commercial activity in the UK. Given the shared interest in images (amongst other things), it is not surprising that the two communities have many connections.
Five significant connections between the two communities are:
Modeling of objects
Virtual environments need realistic objects, and the researchers in the UK are active in two main paradigms. The first is in the use of range sensor data to measure 3D surface points, from which surface models can be created and texture mapped. Groups particularly active in this area are at Edinburgh and Surrey Universities, and 3D Scanners. The second approach processes calibrated or uncalibrated intensity images (based on properties of projective geometry) to register image features in a 3D coordinate system. Oxford and Cambridge University are particularly active in these paradigms.
Modeling of Internal Environments
Virtual environments also need to be created, for navigation, analysis or background texture. Models can be obtained from either texture-mapped surfaces derived from range sensors, or by transforming intensity views of a scene according to the laws of projective geometry. Acquiring models from already existing real environments is an activity at UK Robotics, Oxford and Cambridge Universities, Essex University, Leed University and Edinburgh University's Machine Vision Unit and Virtual Environment Centre.
Embedding humans virtually into real environments requires telepresence, and much previous computer vision experience on, in particular, stereo imaging and servo-driven artificial 'heads' is moving into telepresence research. Research groups at Surrey and Oxford Universities are prominent in this area.
Human Computer Interfacing
Leeds, Oxford and Cambridge Universities, as well as the Olivetti and Canon research labs are investigating how to visually track human hand and head movements, for the purpose of eliminating dataglove and head-tracking interfaces.
Human Motion Capture
Virtual environments are improved by realistic motion of avatars etc. Acquiring human motion via visual information allows larger workspaces and lack of magnetic interference with magnetic trackers. Oxford Metrics and University of Edinburgh are researching this area.
Relevant research at the Machine Vision Unit in the Department of Artificial Intelligence at University of Edinburgh includes:
- automatic acquisition of accurate CAD models of industrial parts
- building models of flexible objects (eg people or mechanisms) using multiple static range images
- a hand-held range scanner for object modeling
- comparison of existing building models to CAD models (just starting)
- CAMERA: automatic acquisition of building models from range data (starting in 1998).
The CAMERA project is a EC-funded TMR network whose main objective is to undertake pre-competitive cross-disciplinary research in the field of automated acquisition of architectural CAD models of already built environments. The CAMERA network consists of: Fraunhofer Institut Graphische Datenverarbeitung (Germany), Instituto Superior Technico (Portugal), Ispra European Commission Joint Research Centre (Italy), Kungliga Tekniska Hogskolan (Sweden), LAAS-CNRS (France), UK Robotics (United Kingdom) and Univ. of Edinburgh (United Kingdom). CAMERA's applicable industrial sectors are chemical and nuclear, training, tourist, museum and archeaological, and construction. The main technologies used are range sensors, computer vision and CAD.
- the British Machine Vision Association: http://peipa.essex.ac.uk/bmva/
- the UK VR-SIG Virtual Reality Special Interest Group: http://www.crg.cs.nott.ac.uk/ukvrsig/
- the Machine Vision Unit: http://vision.dai.ed.ac.uk/
- EdVEC Virtual Environment Centre: http://graphics.ed.ac.uk/EDVEC/index.html
Robert Fisher - University of Edinburgh
Tel: +44 131 650 3098