I have (tried to) express that geometric idea algebraically; yes, geometrically I mean that the residuals are measured orthogonally to the fitted hyperplane (I am not fitting a fancier hypersurface, so the rotation for all residuals is the same, the angle between the "y" axis and the normal to the fitted hyperplane, right?). The formula also seems to match the one (formula (1)) in the Bargiela/Hartley paper, www.doc.ntu.ac.uk/RTTS/Papers/rttg-publ02.ps. If you see an error or know of a different-looking expression for the orthogonal L1 distance, please let me know.
I would also very much like to get more details on the "reweighted L2" implementaion -- it seems close to what the Bargiela/Hartley paper is talking about, which was my question that started the thread. Do you have any references or a description of the method? Thanks, Vitaly [EMAIL PROTECTED] (Gordon Sande) wrote in message news:<[EMAIL PROTECTED]>... > ... Given the rather uncertain terminology in use here, it > might be the case that iterative reweighting with ODRPack will do > your job as at first sight your definition does not match what I > understand by the usual "orthogonal" regression. The usual > "orthogonal" regression has a different rotation for each residual > so the residual is orthogonal to the fitted curve. Maybe that is > what you actually want but have not quite formulated it that way. . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
