[ptx] automatch - some progress picking control points

[Bruno Postle]

On Wed 14-Jan-2004 at 07:16:30AM +0100, Pablo d'Angelo wrote:
> 
> I have started to write a c++ version of it.

Cool, I bet it's faster too (though I have the perl version down to
about 5 seconds for each pair of images).

> I'm using a slightly different
> way of creating candidate triangles:

Looks a bit more intelligent.

> Unfortunately I have some bugs somewhere, so I haven't checked it in yet.
> It creates lots of bogus points.

For any pair of images, there are always many hundred geometrically
equivalent triangles of control points simply by chance.

My approach has been to test lots of triangles; for a pair of images
that actually overlap, particular pairs of points will recur again
and again - It's identifiying and validating these points that is
difficult, simply sorting them by frequency gives good results, but
with some false positives.

My latest version tries to validate each set of control-points by
trying to build them into a network - It has a bug where sometimes
all the points it finds are correct and sometimes all the points are
wrong.

-- 
Bruno