Martin Desruisseaux-2 wrote:
> 
> Olaf Danne a écrit :
>> we still have problems when reprojecting raster data, which we basically
>> already described in 
>> 
>> http://www.nabble.com/Raster-file-reprojection-problems-tf3180801.html#a8826611
> 
> Maybe Simone could tell more. For now, I don't know if the problem happen
> with 
> Geotools GridCoverage renderer or Geotools GridCoverage resample
> operation. 
> Would it be possible to try to reproject a GridCoverage using the
> following code 
> and check if the result is the expected one please?
> 
>     GridCoverage2D coverage = ...;
>     System.out.println(coverage.getEnvelope());
> 
>     CoordinateReferenceSystem targetCRS = ...;
>     coverage = Operations.DEFAULT.resample(coverage, targetCRS);
>     System.out.println(coverage.getEnvelope());
> 
>     ImageIO.write(coverage.getRenderedImage(), "png", outputFile);
> 
>       Martin
> 
> -------------------------------------------------------------------------
> Take Surveys. Earn Cash. Influence the Future of IT
> Join SourceForge.net's Techsay panel and you'll get the chance to share
> your
> opinions on IT & business topics through brief surveys-and earn cash
> http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
> _______________________________________________
> Geotools-devel mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/geotools-devel
> 
> 

Martin,

thanks for your reply. - Simply applying the piece of code you suggested
leads to the strange image 'coverage.png' attached below. It looks the same 
if we use a PolarStereographic or an AzimithalEquidistant projection to
define a target CRS). I guess this is not really what you expected to see
here?

However, I tried to understand a bit more what happens in the
GridCoverageRenderer. I took all the code in an own class and produced all
the 'in-between' images in debug mode using  a PolarStereographic
transformation with latitudeOfOrigin=90N, longitudeOfOrigin=0 and a view
angle of 60deg. The resulting sequence

original.gif
cropped.gif
preScaled.gif
scaled.gif
reprojected.gif
preSymbolizer.gif
final0.gif
ourresult.gif

is attached below.
As you can see, already the 'cropped.gif' is screwed up. At least the
missing part on the west side should not have been cropped. Looking into
GridCoverageRenderer, it can be seen that the cropping is determined by
'destinationEnvelopeInSourceGCCRS' which is calculated in
CRSUtilities.transform. The GeneralEnvelope returned from this method is
[116°33.8'W 41°59.9'S , 180°00.0'E 89°59.5'N] for this case. This is
consistent with the 'cropped.gif' and also the final result. 
When we set this envelope closely to 180 deg. on both west and east side
([170°0'W 41°59.9'S , 170°0'E 89°59.5'N], we obtain the result shown in
'ourresult_170.png'. If I understand correctly the way this envelope is
applied, this suggests to me that we will always loose parts of the coverage
in a circumpolar representation like this, dependent on the longitude
values.  Either parts on the left/right of the image, or triangles close to
+-180deg. are gone.

Am I wrong?  What do you think?

Olaf

coverage.png:
http://www.nabble.com/file/7535/coverage.png 
original:
http://www.nabble.com/file/7528/original.gif 
cropped:
http://www.nabble.com/file/7529/cropped.gif 
prescaled:
http://www.nabble.com/file/7530/prescaled.gif 
scaled:
http://www.nabble.com/file/7531/scaled.gif 
reprojected:
http://www.nabble.com/file/7532/reprojected.gif 
preSymbolizer:
http://www.nabble.com/file/7532/preSymbolizer.gif 
final0:
http://www.nabble.com/file/7532/final0.gif 
ourresult:
http://www.nabble.com/file/7533/ourresult.PNG 
ourresult_170:
http://www.nabble.com/file/7534/ourresult_170.PNG 
-- 
View this message in context: 
http://www.nabble.com/Still-problems-with-reprojecting-raster-data-tf3472146.html#a9735764
Sent from the geotools-devel mailing list archive at Nabble.com.


-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Geotools-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/geotools-devel

Reply via email to