Georeferencing a whole heap of raster imagery

Question

I have a whole heap of satelite imagery coming in that is not georeferenced. They have a grid overlaid so I do know the bounding box/coordinates of the images (and projection). I just need them georeferenced for use in ArcGIS.

I have manually done a few by referencing to a point dataset I traced over the top - and it works great. I just need to automate this process.

Each image is exactly the same, so the same points overlaid will always match up (even if it doesnt, the application of this does not need to be accurate).

Basically I need to apply the 'link table' information from the example I've already done, and apply it to all the images...

Possible in python/model builder?

Answer 1

I'm sure there's more ways of doing this, but we use GDAL's utility program gdal_translate to georeference our PNG images via script on a Linux machine. So, I first retrieved the georeference info from the original data (it's GRIB) used to make the images. Then we set up a script (Linux machine) with that information and used it in gdal_translate's options to georeference all the PNG images in a directory. Works pretty quick for our purposes.

Answer 2

I don't have firsthand experience but I did catch part of the presentation below at a GIS conference in April. Perhaps contact the presenters, might help the brainstorm brew. Or perhaps stir up more ideas here??

Abstract

In Kansas, beginning in the 1850s, the U.S. General Land Office (GLO) commissioned teams of surveyors to conduct transect surveys along all section lines in the state. For each township a plat map was produced that showed forest cover, streams, trails, and other significant features on the landscape, along with corresponding survey notes of additional feature information. This talk will outline the methods that are being used to georeference and digitize forest cover for over 2000 township survey maps for Kansas. In particular, we will focus our presentation on innovative automation procedures that were developed using eCognition, ArcGIS, Python, and MATLAB. This work is made possible through funding from the Kansas Department of Wildlife and Parks - US Fish and Wildlife Service, the State of Kansas GIS Policy Board, KansasView/AmericaView, and the Kansas Biological Survey. Planned and potential applications of these data will also be presented.

1) Integrating GEOBIA and GIS for Automated Georeferencing of 1850s General Land Office Survey Maps (30 Min)

Kevin Dobbs Kansas Biological Survey Lawrence, KS

Other Presenters: Ryan Surface, KARS; Stephen Egbert, KARS; William Busby, KARS;

View Abstract http://www.magicgis.org/magic/symposiums/2012/view_abstract.cfm?pres_id=380

Answer 3

You can use the GDAL tools on Windows to accomplish this with a tiny bit of scripting. You need gdal_translate to add the GCPs to the image and the gdalwarp utility to convert it to a georefererenced image using those GCPs. You can download the OSGeo installer for Windows http://trac.osgeo.org/osgeo4w/ which will give you a linux-like shell (MSYS) with all the GDAL tools.