[fitswcs] 98-05-01 MNRAS paper

[Don Wells]

Regarding polynomials for mapping the geometry of optical imagery, a
relevant paper occurs in the latest issue of MNRAS, which arrived at
NRAO-CV today:

J.E.Morrison, R.L.Smart and L.G.Taff, "Do we need to model plates at
all?", Mon.Not.R.Astron.Soc. 296, 66-76 (1998). 

I still see no reason why a polynomial of sufficiently high order
cannot represent messy geometries with sufficient accuracy. Other
techniques may be used for actual solutions, and the simple polynomial
be used only for interchange, of course. The numerical maps shown in
the paper would be well-represented by 2-D splines and probably also
well-represented by high-order 2-D Chebyshevs. Splines and Chebyshevs
can always be converted to simple 2-D polynomials.

-Don