Good to know that the order matters. But still, "thin plate" is not what you 
want. "thin plate" is for maps that have local micro distortions that can be 
corrected the same way as you would do with a foil that has reference points 
and you force these points to match. 

Usually a normal scan has an error in scale, offset and rotation. For these 3 
properties, 3 points are sufficient to resolve the linear equation system. 

If the map has a grid you can supply more points and quadratic interpolation 
is possible. For scanned maps this is the best.

Am Sonntag, 3. September 2017, 15:22:34 CEST schrieb solitone:
> On Sunday, 3 September 2017 10:28:30 CEST Oliver Eichler wrote:
> > Do not use thin plate. It's pretty special. With 4 points you have to use
> > linear. There is no more information in 4 points than a linear
> > interpolation.
> 
> On the menu entry, linear is associated with 3 points. I tried it with the
> first 3 points, and it worked. Then I re-tried with 4 points, using thin
> plate, but swapping the 3rd and 4th point, so that they are ordered
> clockwise on the plane. With this change, also thin plate worked.
> 
> Here is what I mean:
> 
> Before:
> 
> ----------------------------------------------------------------------------
> - -gcp 463 793 320000 4978000
> -gcp 4273 347 328000 4979000
> -gcp 441 3639 320000 4972000
> -gcp 4726 3196 329000 4973000
> ----------------------------------------------------------------------------
> -
> 
> After:
> 
> ----------------------------------------------------------------------------
> - -gcp 463 793 320000 4978000
> -gcp 4273 347 328000 4979000
> -gcp 4726 3196 329000 4973000
> -gcp 441 3639 320000 4972000
> ----------------------------------------------------------------------------
> -
> 
> Cheers,
>   Davide



------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
Qlandkartegt-users mailing list
Qlandkartegt-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/qlandkartegt-users

Reply via email to