wxwindows experiments

[Pablo d'Angelo]

On Mon, 19 May 2003, Juha Helminen wrote:

> Strange compiler messages will not be problem if code do not have too 
> much gimmicks or compiler dependent features.

The problem with wxWindows is that the macros are an integral part,
and they are needed if one does not want to create applications
in the monolithc everybody knows everything way. The events depend on many
macros, and events are the glue to compose otherwise independant widgets
together.
Macros are also needed for the rtti system of wxwindows.

> What about making 
> search algorithm so that it do not need to calculate every pixel 
> position... Try to find local optimum points and leave out (/no 
> calculation) propably bad locations.

I'm already doing that (sort of), because correlation is very
computationally itensive. My control point find code does correlation
on low resolution images first, and then progressively looks at the
better hits in higher resolution images. Else it would be terribly slow.
The approach can probably be optimized a bit more by deciding the
number and size of these resolution steps based on the template and
image size

Ok, I made the decision to support the wxwindows development.

ciao
  Pablo
--
http://wurm.wohnheim.uni-ulm.de/~redman/
Please use PGP