I have been involved in remote sensing for some years now during studies. I am wondering which combination of software and image processing modules would be most suitable for an entire process chain for professional image processing solutions and products. I am very curious about what other users think and perhaps it will even lead into an interesting discussion from experienced users.

What I mean in detail is the combination of:

  • a database storing geographical data like shapefiles but especially huge amounts of satellite imagery with its corresponding metadata

  • image processing modules automatically using appropriate data from the database useful for the chosen processing step (e.g. all satellite data with at least a certain spatial resolution for a given time span and geographic area to produce change detection maps)

  • the results then should be integrated into the database and perhaps even be available for distribution via a web server

Unfortunately I do not have any advanced knowledge on databases for geographic data. Perhaps GeoNetwork/GeoServer with PostGIS would be an option?

For the image processing modules I thought of either implementing necessary algorithms in C++/GDAL or JAVA/Geotools. Also there would need to be some kind of module connecting to the database in order to fetch the needed data for processing and the creation of metadata for the processed images/products. My thoughts were that generally open source solutions would be best as such a system would be developed for a long run and being independent of commercial companies would be desired.


you may check out OTB[1] and OSSIM[2] and ILWIS(only for Windows). All are open source tools.

There are commercial tools used in RS such as ENVI, LCCS, ERDAS, Leica LPS(mainly photogrammetry) now part of Integraph)

[1] http://ossim.org

[2] http://www.orfeo-toolbox.org/otb/

| improve this answer | |

You might want to develop something as an add-on to Qgis, and connect to a PostGIS database.

Qgis is an open-source desktop GIS with an embedded python interpreter and access to image processing toolkits. You can develop plugins that use the Qt user interface toolkit and have access to Qgis internals via the Qgis-Python API.

If the standard image processing toolkits aren't enough, you can also write plugins in C++ - but you might find that reading rasters into numpy arrays in python and working with them is fast enough.

| improve this answer | |
  • I know Q-GIS well but I only use it for simple editing and data viewing as for processing I use different RemoteSensing software. I would actually think that when processing large amounts of e.g. Landsat data the speed difference between Python and C++ would be quite significant. Also I though having 'standalone' image processing modules would be more powerful e.g. for setting up process chains. A question to PostGIS: Is it actually suitable for saving huge amount of raster data? Is this embedded into the database or are the files still on harddisk. – DomR Sep 24 '12 at 7:37

I've not written any modules, but have used some user compiled routines (IDL) in ENVI. We also had someone write a module for us. Seems to be quite powerful and flexible, though of course you need an ENVI license to run them.

| improve this answer | |

You can try out raster management options of ArcGIS Desktop [1]. You would have to pay special attention if you do not want to modify your pixel values or want to store data in native remote sensing formats.

You can also checkout Envi tools for ArcGIS [2]. Envi is an established remote sensing image processing package.

[1]. http://resources.arcgis.com/en/help/main/10.2/index.html#/Design_methodology_for_a_raster_database/009t0000002w000000/

[2] http://www.exelisvis.com/Learn/WhitepapersDetail/TabId/802/ArtMID/2627/ArticleID/9895/ENVI-Tools-for-ArcGIS174-and-ENVI-for-ArcGIS174-Server.aspx

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.