Recent Changes - Search:


edit SideBar


Estimation Of Stereo And Motion Parameters Using A Variational Principle

H. P. Trivedi, Image and Vision Computing, 5, 2 ,pp.181-183 May 1987.

Many papers have been published over the years on the subject of Stereo camera calibration. In particular there have been many publications on the subject of uncalibrated stereo, and estimation of the fundamental matrix. In fact the statistical basis for the closed form numerical approaches adopted in these techniques is quite often shaky, and the degree of freedom permitted by allowing non-orthogonal sensor axes is mathematically impressive but unnecessary. This early paper on the subject of automatic calibration used a valid statistical principle for the determination of parameters of a stereo camera system for an orthogonal sensor matrix. It is not closed form, but it minimises something like the right thing, and can also be extended to use robust kernels and radial distortion terms. It has all of the scope for application of uncalibrated techniques but, for the cost of a few side of the box sensor calibration terms, produces a Euclidean geometry with one overall scale factor. Compare this with the better known techniques which do not minimise a valid expression, cannot deal with radial distortion, and leave an un-resolved 4x4 homology matrix.

NAT 8/3/2005

Edit - History - Print - Recent Changes - Search
Page last modified on March 05, 2012, at 03:55 PM