[gdal-dev] [RFC] [GDAL] Idea for GSoC, 2014

Even Rouault even.rouault at mines-paris.org
Tue Jan 28 12:01:00 PST 2014

Le mardi 28 janvier 2014 20:41:15, Kshitij Kansal a écrit :
> Hello
> Thank you for you suggestions.
> Joaquim
> I looked into the link regarding Affine-SIFT that you provided. Indeed the
> results that are shown there are very promising. If we can bring this to
> one of the core functionality of GDAL then it would not only help in this
> project but many more digital image/computer vision related projects.
> Rutger
> Yes the main aim to propose this idea into GDAL community is to bring this
> functionality into the core GDAL functionality. I checked with the other
> communities like QGIS or GRASS or even OPTICKS, but this functionality of
> automatic geo-referencing is not present and till now no work has been done
> in this direction. If implemented here, these communities can directly
> adapt our thing and then build on this.
> As we can see that we have a lot of algorithms to start with. Some kind of
> analysis like "Estimating the shift between two georeferenced images". This
> was something Chaitanya was talking about.
> Also there was something me and Chaitanya discussed about, like matching
>  raster with vector datasets, for example, GPX data could be used to match
> roads.
> Once we have this available in our core functionality of GDAL, we can
> easily use it to make "*Raster / Vector Geo-referencer on the Web*" which
> was the original idea proposed by the community in GS0C'13.
> Any Comments or Suggestions?


I just wanted to mention, to avoid any potential problem later in the process, 
that if you adapt existing code from other projects into GDAL, you will need 
to make sure that it comes with a license compatible with the GDAL X/MIT 
license (so LGPL or GPL would not be OK), or make sure with copyright holders 
that it is OK to port it to GDAL under the GDAL X/MIT license.

As a result of last year GSoC, we have an implementation of SURF algorithm in 
GDAL. Not sure how mature it is, but that might be a potential starting point.


Geospatial professional services

More information about the gdal-dev mailing list