Thanks Jukka, that is an interesting article in its own right, and it is
relevant to historical maps too. Your solution is viable, but it takes
more work (insert corner gcp in each image), which shouldn't be
necessary, as this information is already available but is discarded by
gdal_translate. If gdal_translate could keep this information in the
target file, even when gcps have been added, everything would be fine
(and my impression is it should).
Jan
Rahkonen Jukka wrote:
Jan Hartmann <j.l.h.hartmann <at> uva.nl> writes:
Jukka Rahkonen wrote:
>> Hi,
>>
>> I am not sure if I understood how you trying to do the job, but
hopefully my
>> task and work flow are close enough to be useful for you.
>>
>> 1. Let's say that the original image is in epsg:2393 and target is
epsg:3067.
>> 2. I insert some ground control points in epsg:3067 with
gdal_translate and
>> -gcp. Image format of the temporal output file can be GeoTIFF but
also virtual
>> raster(vrt).
>> 3. I warp the temporary image, now containing gpc's in epsg:3067
projection, to
>> final product using gdalwarp as
>> gdalwarp -s_srs epsg:3067 -t_srs epsg:3067.
>
>
>> Hi Jukka,
>> You have described exactly what I want to do, but the problem is that
>> in step 2, inserting gcps, the georeferenced boundaries of the file are
>> lost. Eventually you get a somewhat displaced raw raster file.
>> Jan
Hi,
That displacement could be avoided at least by inserting manually gcp
to each
corner of the image. Perhaps this article gives some ideas:
http://www.scangis.org/scangis2007/papers/r3_rahkonen.pdf
-Jukka-
_______________________________________________
gdal-dev mailing list
gdal-dev@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/gdal-dev
