[GRASSLIST:2318] Re: Question about import of non georeferenced
rshepard at appl-ecosys.com
Wed Aug 8 09:53:17 EDT 2001
On Wed, 8 Aug 2001, Moritz Lennert wrote:
> If I'm not mistaking, a scanned image can be imported into an x,y-location
> which has an x,y-"region". It is just not projected geographically, nor
> I'm not a specialist in image treatment, but the way I've always understood
> it, if you do not have any more information than the basic x,y resolution
> of the file, you import it into an x,y location, then leave Grass, restart
> Grass with a location in the projection you want, then use i.rectify (in
> combination with i.target, i.points, etc) to import/georeference the raster
> image into your projected location.
And this is correct, too. My point -- not clearly made -- is that the user
needs a well-distributed set of geographic locations, measured on the
ground, for points that can be readily identified on the scanned image.
Without such points the scanned image cannot be rubber-sheeted and
georeferenced. That's all I was trying to communicate. And, getting that
information is not always easy.
We've done this quite a few times with our high-precision (+/- 30 cm) GPS
receiver. It's easy to acquire positions at points on the ground (say, the
center of a road intersection) but not that easy to identify that point on a
scanned aerial photograph when trees cover one edge of the road. Or, when
the area of the photo is out in the desert and there are no roads, built
features or other points on the photo that can be located on the ground.
So, conceptually the scanned image can easily be georeferenced in GRASS.
Getting sufficient data over a large enough area can be a problem.
Dr. Richard B. Shepard, President
Applied Ecosystem Services, Inc. (TM)
2404 SW 22nd Street | Troutdale, OR 97060-1247 | U.S.A.
+ 1 503-667-4517 (voice) | + 1 503-667-8863 (fax) | rshepard at appl-ecosys.com
Making environmentally-responsible mining happen.
More information about the grass-user