Leonard Coop coopl at ava.bcc.orst.edu
Sun Jul 2 08:00:00 EDT 1995

On Sat, 1 Jul 1995, BARNES DAVID wrote:

> Dear Grass-users,
>       I don't suppose many U.S. Grass users are out there on this weekend 
> but I'm pressing on a project and having trouble with i.points.
>       I attempted to i.rectify some air photos using Topo quad determined
> UTM coordinates in i.points. Seemed to work O.K. for image #1 with col and
> row errors < 20 units (actually image x,y coords are equal to pixels on input
> image, ~ 1m) and target error <10. The overall rms error <10m, less than TOPO
> sheet resolution -> I'm happy.
>       The next image (#2) was processed in the same fashion. This time many
> of the col and row errors are over 100 image units but the target errors are
> still <10. Both sets of image points in the two i.points exercises (#1 and
> #2) were taken from about the same places on the respective air photo so that
> photo distortion should be similar and not that bad (not hundreds of feet!).
> A third image was similarly processed with the same (bad) results as #2. 
>       Both unsatisfactory rectified target images (#2 and #3) produced by
> i.rectify misrepresent the input location reference points in the UTM
> coordinate system as determined in d.measure "where am i". The locations of
> reference points input for image #1 are O.K. in the rectified image.
>       The last puzzle is that the two incoorectly rectified target
> rastermap images (#2 and #3) do not display the entire input rastermap images
> dispite instructions to set default region to be "minimum to contain entire
> image" in the i.rectify interactive commands. Any ideas - wise advice?

1) try 2nd order rectification using i.rectify2; use as regular a
   spread of points as possible, as many as is practical (25 points
   for 2nd order is not too many).
2) try setting the exact boundaries and resolution you want in the
   target location, I have found it works much better than letting
   it set the region as you have been doing.

Len Coop       coopl at bcc.orst.edu

More information about the grass-user mailing list