[Kstars-devel] Extending Hendry Draper numbers in Kstars

[Alexander Potashev]

Hi,

2015-06-23 13:25 GMT+03:00 M.S. Adityan <msadityan at gmail.com>:
> Hi Alexander,
>
> How would choosing arbitrary points on the image help? The idea is to
> display the constellation image exactly as it is described in ancient maps
> and atlases of the sky. This means suppose the head of Leo falls on Star X,
> then that is how it should appear in KStars. This means the Star X acts as
> an anchor for the head of Leo. Hence I cannot choose arbitrary anchor
> points, but that anchor points should be real stars.

Imagine you have 3 stars (or may be 2, see below) and their image
coordinates (in pixels) and RA/DEC coordinates (e.g. in radians.)
These coordinates define a map from the image to a spherical surface,
i.e. from these you can determine RA/DEC coordinates of arbitrary
points in the image.

Now you can take some other 3 points in the image, say (0 px, 0 px),
(0 px, 1000 px) and (1000 px, 0 px), and apply the above defined map
to them. You will have 3 points again with their image and RA/DEC
coordinates. These 3 points also define a map from image to the
scherical surface, like it was for the 3 stars. When you apply the new
map (induced by coordinate pairs for 3 points) to the 3 stars, their
image coordinates will map to their _correct_ RA/DEC coordinates, just
because the process of generating a map from the 3 coordinate pairs is
deterministic.

So, no -- the anchors don't have to be real stars.

> Yes, now I am trying to do this with RA/DEC coordinates of stars in the
> J2000 system. I am also considering just using 2 points to simplify the
> math, because I can always preserve the aspect ratio of the image, and hence
> I would only require 2 anchor points in total to scale/rotate the image. I
> am doing this for 1 constellation right now. Till then hopefully we should
> figure out what requires to be done with those HD numbers which are unknown
> to KStars. Either the HD number list in KStars is extended, or I make a new
> list of anchor stars ALL of which are known to KStars. That would be quite a
> lot of work, but will have to be done.

2 points are enough to define a map from a flat texture to a spherical
surface. But I'm not sure if we can assume the image to have the same
angular resolution on both axes (what you called aspect ratio).

-- 
Alexander Potashev