From: ian@robots.ox.ac.uk (Ian Reid) Subject: Re: MEDIA: "Virtual reality is bad for your health," says BBC. Date: Wed, 2 Aug 1995 09:11:12 GMT Message-ID: <1995Aug2.091112.24332@thomas.robots.ox.ac.uk> Organization: Robotics Research Group, Engineering Science Dept, Oxford, UK. In article <139564294wnr@sirapgc.demon.co.uk> Chris Seal writes: Agreed, it was fairly pointless as an end in itself (doesn't change a thing about the decision) but you wouldn't know it from the media coverage. However it was a very useful demonstration of the power of *uncalibrated* techniques. It is precisely through the use of these methods that the factors you mention above are irrelevant. -- location of cameras are unknown but unneeded; -- lens characteristics such as focal length are unknown and unneeded; -- camera resolution is unknown and unneeded; -- lens radial distortion can be measured from the image using known straight features (ie goal lines etc) and shown negligible; -- simultaneity of views can be verified in a straightforward manner; -- calibration required for metric measurements (ie where the ball landed) can be derived directly from known lengths in the world (eg size of 6 yard area). While the 1966 goal was a rather frivolous demonstration, it shows that useful, *accurate* information can be determined *without* knowing what camera(s)/lens(es) captured the images. A consequence of this is that useful tasks such as accident reconstruction, computer enhanced imagery for teleoperations, and creation of virtual worlds from (say) hand-held camcorder tapes will be possible in the near future. I have a web page which explains briefly what was done in the '66 world cup instance: http://www.robots.ox.ac.uk:5000/~ian/wc66 and for links to papers describing work on uncalibrated vision for mobile robotics and surveillance, see http://www.robots.ox.ac.uk:5000/~lav and http://www.robots.ox.ac.uk:5000/~ian Cheers, Ian -------------------------------------------------------------------------- Ian Reid _--_|\ Dept of Engineering Science / \ Oxford University, OX1 3PJ \_.--._/ ian@robots.oxford.ac.uk (01865) 273168 v http://www.robots.ox.ac.uk:5000/~ian