[gdal-dev] extract vector/raster data from GeoPDF

Joaquim Luis jluis at ualg.pt
Thu Sep 3 10:18:52 EDT 2009


I was also digging a bit on this matter. It's true that the blogger only opens
our appetite as there is no more details on how to do it generally. I mean, for
UTM maps it has become fairly obvious how one can do it. The relevant part in the
PS file is

[ {ThisPage} <<
/LGIDict <<
   /Description Title
   /CTM [(35.28267) (0) (0) (35.28267) (205188.64) (3207094.8)]
   /Projection <<
     /Description (WGS 84 UTM 16N)
     /ProjectionType (TC)
     /Datum (WGE)
     /CentralMeridian (-87.0)
     /OriginLatitude (0.0)
     /FalseEasting (500000.0)
     /FalseNorthing (0.0)
     /ScaleFactor (0.999600)
     /Type /Projection
    /Type /LGIDict
    /Version (2.1)
 >> /PUT pdfmark

However, there is no more info on how to program in this "TerraGo way"
A next post shads a bit more lite on the matter.
As you can see in the quada.ps file there is an alternative "Adobe way",
which encodes the referencing as

% embed georegistation info
[ {ThisPage} <<
   /VP [ <<
     /Type /Viewport
     /BBox[0 0 dh dv]
     /Name Title
     /Measure <<
       /Type /Measure
       /Subtype /GEO
       /Bounds[0 0 0 1 1 1 1 0]
       /GPTS[29.0 -90.0 30.0 -90.0 30.0 -89.0 29.0 -89.0]
       /LPTS[0.024324 0.039461 0.051689 0.978774 0.975675 0.961136 0.957432 0.022122]
       /GCS <<
         /Type /PROJCS
 >> /PUT pdfmark

Now this is more familiar, except for the
/LPTS[0.024324 0.039461 0.051689 0.978774 0.975675 0.961136 0.957432 0.022122]
part that, contrary to the promise has not been explained where it comes from.

I think I could implement the first encoding type in GMT's ps2raster (if I knew all the specs)
Second type would be even easier ... if we had a way to translate the GMT projection syntax into
WKT, which we don't.


> Joaquim,
>   Doing the format conversion from .ps to pdf is one thing (and there 
> are several ways to do it), but embedding the georeferencing in the PDF 
> to make it a GeoPDF is the interesting bit.  I haven't found any open 
> source project capable of do that.  Or even capable of writing the PDF 
> elements (e.g. Frames) necessary for writing the GeoPDF information.  
> But I remain hopeful...
> Brent
> Joaquim Luis wrote:
>> Brent Fraser wrote:
>>> I see the example calculation of the transformation matrix, and a 
>>> statement "I created a GeoPDF by running the Postscript file through 
>>> Ghostscript to create a Postscript file that looks like this"
>>> But I don't see where the georeferencing is written into PDF to make 
>>> it a GeoPDF.  Is it a Ghostscript command line?
>> Brent,
>> One only has to convert it to pdf using ghostscript.
>> I did it with GMT's ps2raster, like that
>> ps2raster quad.ps -Tf -A
>> Joaquim Luis
>>> Brent
>>> Klokan Petr Přidal wrote:
>>>> Hi,
>>>> There is a great blog post (and the linked "worked example" post 
>>>> with details):
>>>> http://geopdf.blogspot.com/2009/02/geopdf-map-for-worked-example.html
>>>> It shows you how to create geopdf via GhostScript - so there is
>>>> already a practical open-source example how to encode the georeference
>>>> into the PDF/PS according the OGC standard - for use in Acrobat
>>>> Reader. To add support for such tag in MapServer, which generates pdf
>>>> dynamicaly via pdflib, should not be totally problematic.
>>>> Decoding is not as hard either, there are nice libraries like poppler
>>>> (http://poppler.freedesktop.org/), which allows you to parse vectors
>>>> (and convert them to SVG for example) or rasterize the PDF files (into
>>>> TIFF,...) via Cairo.
>>>> The work is in assigning correct geographic coordinates to the
>>>> coordinate system internally used in PDF files and especially write
>>>> the bridge to the outside world (with GDAL/OGR).
>>>> I am afraid that authors of the GeoPDF standard would not like this,
>>>> as it seems that the idea of GeoPDF is "see it in the Acrobat, print
>>>> it, but that's all". At least I think so, because they discontinued
>>>> their Geopdf2geotiff product and all the conversion tools are just one
>>>> way - into GeoPDF. Please correct me...
>>>> Anyway, in this moment you can quite easily use utility like
>>>> "pdfimages" to extract full quality image tiles from any GeoPDF (like
>>>> those from USGS) and merge it based on their location in PDF into one
>>>> GDAL file via VRT (gdalbuildvrt) with a bit of hacking. This is what I
>>>> did for my favorite USGS DRG of Grand Canyon ;-).
>>>> Look at: http://klokan.mzk.cz/~klokan/geopdf/ - soon I will update the
>>>> MapTiler.org overlay examples...
>>>> Unfortunately all PDF parsing libraries I know are GPL, and that means
>>>> we can't use them for the gdal driver - because of the license issues.
>>>> But to create a GPL utility for converting GeoPDF to anything what
>>>> GDAL/OGR supports should be OK. Poppler can be the best base of such
>>>> GDAL-based utility for reading/rasterizing of the GeoPDF files.
>>>> Now just find a sponsor and time to make it ;-).
>>>> Best,
>>>> Klokan Petr Pridal
>>> _______________________________________________
>>> gdal-dev mailing list
>>> gdal-dev at lists.osgeo.org
>>> http://lists.osgeo.org/mailman/listinfo/gdal-dev

More information about the gdal-dev mailing list