CCD's are widely used for astrometry. Since a spherical surface (the celestial sphere) is projected onto a rectangular surface, the plate scale or pixel scale (approximated with 1/focal length etc) is used to estimate the angular separation of stars in an image.
However, due to the actual plate scale being non-uniform across the whole sensor, different images with the same constellation of stars at different positions on the sensor will result in distorted images, compared to a centered image of the stars.
Hence for accurate work e.g. astrometry,
Why are pixel scales usually assumed to be constant? Is it because other sources of error would usually dominate hence accounting for differences in pixel scale is unnecessary?
For astrometry, how are these geometrical distortions corrected?
Is there any systematic way to read up on this topic? (of optics, imaging etc)
What I've found so far:
The series of articles:
Ground-based CCD astrometry with wide field imagers
(J. Anderson et al, 2006) https://doi.org/10.1051/0004-6361:20065004
(auto-calibration described in detail)
(A.Bellini and L.R.Bedin, 2010) https://doi.org/10.1051/0004-6361/200913783
(here the authors discuss auto-calibration briefly.)
Since the scale is a free parameter in deriving GD correction, choosing a particular scale value will not invalidate the solution itself.
Not sure whether the above remarks in the second paper means that non-uniform pixel-scale is not a concern, at least to the algorithm.
And a book Stellar Paths Photographic Astrometry with Long-Focus Instruments, Peter Van De Kamp
Plate solving is also discussed in Spherical Astronomy, Smart