Looking for a good example to teach image processing on GPU

Dear all,

we are planning a GPGPU course at our university and I would like to contribute a small example about image processing.

The course will not focus on image processing as such, but rather on how to program for GPGPU and its advantages compared to CPU.

I am now looking for a nice, not-too-trivial example that is at the same time not overly complex. Writing a convolution and/or doing a FFT or something like this seems adequate, but at the same time a bit too simplistic, because it is something that one would usually use a ready-made library for.
I would therefore be glad to hear other suggestions. Any resources would be nice too of course.

What I am targeting for is an example with:

  • ~250 lines of code
  • shows how to program a kernel for image processing in a not too complicated way
  • exhibits the performance gain over a similarly (non-)optimized CPU - by “similarly (non-)optimized” I mean, that I do not want to focus much on optimization aspects

I would be grateful for any suggestions.

Thanks and best regards,
Michael