I have a set of aerial images which I am trying to clip by their overlapping areas, for use in a more involved program I am writing in Python with Python Imaging Library (PIL).

Is there a command line based tool out there with an easy method for extracting the overlapping areas from a pair of images?

I know this may be difficult since the images will have slight perspective differences, but I was hoping it would still be possible.

I was thinking a library such as GDAL may have this sort of capability but I am unfamiliar.

  • 1
    Do they have georeferencing? i.e. when you view them in your GIS package do they fall in the correct location with respect to existing features? Jul 28, 2014 at 2:09
  • Unfortunately no georeferencing information is included with this set of images. When loaded into ENVI and QGIS, they're all just stacked on top of each other. They're 8-bit jpegs so I'm assuming this information was stripped from them or lost at some point. That being said, it appears the only way to align these would be visually... sadly.
    – mryan13
    Jul 28, 2014 at 19:41
  • You could try add basic georeference on these images, it will take time and not be perfect as stereo pairs aren't orthorectified. In this case I think something is better than nothing. Use only affine transformation (scale and rotate) so as not to distort the images like the higher order transformations. If you can save the transformation and apply it to a box in pixel coordinates to create an image mask and then overlap the masks. Jul 28, 2014 at 21:47
  • Thanks for the input! Affine transformation sounds like what I would want, because I want to line the images up as close as possible based on scene content, while maintaining the perspective differences from parallax.
    – mryan13
    Jul 29, 2014 at 22:43
  • Michael, I've just found a shapefile for the images which does include a bounding box for each image collected. This might be useful! Reposted the question at gis.stackexchange.com/questions/108943/…
    – mryan13
    Jul 30, 2014 at 21:56


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Browse other questions tagged or ask your own question.