# How to do geometric correction of remote sensing data in R?

My challenge is to replace most of the remote sensing commercial tools with R code (I'm teaching assistant at the University and I would like to move to open source R code to show the added values to the students). I'm currently stuck as I cannot find a way to perform geometric correction of remote sensing data. The exercise should be:

1. Select Ground Control Points on both the georeferenced image and the image to correct
2. Use polynomial mapping for image correction (students should try to test different polynomial order and calculate the RMSE for each step)
3. Warp the image in a defined projection using the interpolation method (either nearest neighbour, bi-linear interpolation, cubic convolution)

The 2 last steps could be processed using GDAL but it is not really interactive with R.

Does someone has any idea or experience with such processing?

-
+1 Interesting topic. However, you will find R to be an unsuitable platform for image processing in the remote sensing world. – Aaron Dec 2 '13 at 20:48
@ Aaron : I agree with you for some points. But for image processing such as matrix algebra, it appaered to be really powerfulll and easy to use. – Wraf Dec 4 '13 at 8:23
@Wraf Take a look at the gdalUtils package on r-forge (you can do both 1-2 with this package) and you can use the raster package for the third point. Another interesting open source project is GRASS (R integrated through the spgrass6 package) – dickoa Dec 11 '13 at 14:21

My experience when working at the Remote Sensing Group at the Centre for Geophysical Research, University of Costa Rica was quite satisfying. Actually I designed Image Processing software at the user level and other members of the group who were proficient programmers wrote the routines using C++, Turbo Pascal and assembler.

The routine for geometric correction using affine transformation was written by Dr.Javier Soley, who died on 25 December 2012. Anyway, the routine runs on a 32 bit MS-Windows platform from the system prompt. You will have to rewrite the routine in order to fit your programming environment and I am willing to help you.

Please do review the paper `Remote Sensing at the Centre for Geophysical Research, University of Costa Rica.` J. A. Amador, A. S. Burgos, F. J. Soley, G. Serpas and A. Vargas. "International Journal of Remote Sensing.", 1993, Vol. 14, No. 12, 2391 2401.

A. S. Burgos, burgosalvaro@gmail.com

-
Welcome to GIS SE! Our protocols can take some getting used to but one of them is that we do not include signatures (contact details) in any Questions or Answers because every post we make is already signed by our user cards. Your user card is located here so I recommend you just move your name and email address from this Answer into that. – PolyGeo Jan 3 '14 at 2:06
@PolyGeo: You might be right, but I think this answer could help the asker a lot, using mail exchange for further assistance. For others, a printed reference is given. – AndreJ Jan 3 '14 at 9:30
@AndreJoost I'm only suggesting that the name and email address (not the Answer) be moved as per the help guidelines because I agree that the reference and offer of assistance looks very useful to the asker (and later readers). – PolyGeo Jan 3 '14 at 9:59
The email address in the user card can not be seen by others. This is useful in most cases, but not here. You can remove it if contact is established. – AndreJ Jan 3 '14 at 11:27