Leonard Coop writes:
I have scanned a map of the western US (lambert projection)
and want to register it as lat-long.i.rectify does a poor job. It looks OK at the edges or in the middle
points but not both. The error is as much as 200 miles. The number of
points I use does not affect error very much. Is the problem
because i.rectify expects a more planar image (such
as a satellite image)? Do I need a way to convert the lambert
(conical) projection into a cylindrical one? Grass 4.0 does not seem to
have a module to convert coordinate systems in this way. How do we
get a more accurate registration/rectification?This must be a problem solved every day somewhere, but we are stumped for
now.
You raise several interesting points. What you really want to do is a projection
transformation, and i.rectify is not capable of this. If you were working with
vector data, you could use v.proj to do the transformation. If what you are
really after is linework off of the scanned map, I would suggest transforming it
to vector *before* bringing it into GRASS (LTPlus can do this, if your scanning
software can't).
However, if you are really after raster data, you will need to "rubber-sheet"
your data. In GRASS, your only option is to use i.rectify2 in GRASS4.1 (another
good reason to get 4.1!!), which can do 2nd- and 3rd-order transformations.
I just remembered something else. There was a posting on the grassu-list a
couple of weeks ago regarding a contributed program called r.proj . If you
didn't save this, I can forward it to you (the e-mail, not the program!).
Regards,
--
Malcolm D. Williamson - Research Assistant E-mail: malcolm@cast.uark.edu
Center for Advanced Spatial Technologies Telephone: (501) 575-6159
Ozark Rm. 12 Fax: (501) 575-3846
University of Arkansas
Fayetteville, AR 72701