Re: [Gpsbabel-misc] Averaging tracks
Brought to you by:
robertl
From: Kacper P. <kac...@gm...> - 2011-02-18 15:30:29
|
On 18 February 2011 13:37, Jesus Cea <jc...@jc...> wrote: > [...] > Could you possibly describe your algorithm?. As I said I'd like to implement the solution in PERL. But prior to getting ready algorithm would try to summarise the assumptions. - we have a "clod" of points and would like to find a curve that "best fits" the "cloud" - point lay on the sphere and any one may be described by two dimensions - above affects the mentioned curve - we do not consider the error in measuring the points coordinates My weak memory from academic background let me suppose that it is well to describe the problem on the two dimensional metric space. Which leads to the questions if simple distance on a flat (Euclid's) area with the Cartesian coordinates is enough for us, or we should consider distance in geodetic terms as the part of of a "big circle". I'm not sure if a "big circle" is a proper geodetic therm in English language. Other echos of mathematics tell that Bezier curves may be a good approach. Generally Galerkin method should produce the proper fit. I haven't studied such a things for about 15 years so may mistake. Would await some discussion that lead us to the proper solution. KAcper |