Radarsat-2: Current version of GDAL doesn't handle GCPs in the correct pixel position

[sbfitzpatrick]

The radarsat-2 imagery raw file can be directly opened by using the GTiff driver or the product.xml can loaded via rs2 driver. In case of loading a product.xml file the metadata are pulled fomr the product.xml and doesn't take GCPs from the raw image. The interpretation of these is offset by a half pixel. This change addresses problem.

[Jukka Rahkonen]

Sounds somehow related to pixel is point/area thing.

RFC 33 http://trac.osgeo.org/gdal/wiki/rfc33_gtiff_pixelispoint was implemented in 11/23/2010 in http://trac.osgeo.org/gdal/changeset/21158.

I am not sure how to understand the following comment and code in r21158. Do they really suit together, or should I understand, that rs2dataset is making half a pixel shift for "pixel is area", while r21158 if doing is for "pixel_is_point"

Comment from rs2dataser.cpp 16/10/2010: pixel/line coordinates updated to add 0.5,0.5. According to new spec update, GCP with pixel/line 0,0 actually corresponds to the pixel extending from (0,0) to (1,1), so a pixel/line shift of 0.5 is needed for pixel-is-area interpretation (this will make the product.xml GCPs consistent with those of the corresponding pixel-is-area Geotiff for a given product).

code from r21158

if( bPixelIsPoint && !bPointGeoIgnore ) {

adfTiePoints[3] += adfGeoTransform[1] * 0.5 + adfGeoTransform[2] * 0.5; adfTiePoints[4] += adfGeoTransform[4] * 0.5 + adfGeoTransform[5] * 0.5;

}

[Even Rouault]

sbfitzpatrick,

would you mind attaching the gdalinfo report on the .tif itself ? It would be interesting to see if it is reported as AREA_OR_POINT=Area or AREA_OR_POINT=Point. I'm wondering if the half pixel shift in the RS2 driver must be done systematically, or depending on the value of AREA_OR_POINT in the .tif : is it clear in the RS2 spec ?

[Jukka Rahkonen]

sbfitzpatrick,

Still waiting for the gdalinfo report.

[Even Rouault]

This ticket has been automatically closed because Trac is no longer used for GDAL bug tracking, since the project has migrated to GitHub. If you believe this ticket is still valid, you may file it to ​https://github.com/OSGeo/gdal/issues if it is not already reported there.