A small standalone tool to crop satellite images and their RPC.

Suppose you have two images im1.tif and im2.tif, and the associated RPC coefficients in the files rpc1.xml and rpc2.xml. The call

$ ./rpc_crop.py out im1.tif rpc1.xml im2.tif rpc2.xml 12300 4200 800 600

will write in the folder out the cropped files. The cropped region is defined in the first image by the last four arguments. In the example above, 12300 4200 are the column and the row of the top-left corner of the region, and 800 600 are the width and the height of the region. The second image is cropped in such a way that the region visible in the first crop is still visible in the second crop. This is done using SRTM data.

To interactively select the region to crop, use the rpc_crop_gui.py script. A preview of the reference image is needed:

$ ./rpc_crop_gui.py outdir preview.jpg im1.tif rpc1.xml im2.tif rpc2.xml

Nothing to do, the only needed binary will be compiled on the fly during the first call to one of the python scripts.

• python 2.7 or 2.6
• numpy
• pyopengl
• gdal
• g++
• libtiff

On Ubuntu 13.10, all the needed dependencies can be installed through the package manager:

$ sudo apt-get install python-opengl python-numpy gdal-bin libtiff4-dev g++

Note that the use of SRTM data requires an internet connection, at least the first time you run this tool on a dataset.

Gabriele Facciolo, Enric Meinhardt-Llopis, Carlo de Franchis