I started this project few months ago. Now I'm going to ressurect it. cookie_generator.py is a script I wrote then. I plan to use the same idea, but implement it in different way. 

Specification:
Input:
an image
how should it look like?
image or picture should be considerably simple and appriopriate, so generating cookie_cutter from it would be possible
Operations:
first - image preparation, smoothing, grey-scale, something like that
I would like to do some corner detection. I've already seen some nice code in a book about computer vision in python.
Processing points to parameters needed to generate openscad script.
Output:
openscad file, which would be a model of a cookie cutter

Brilliant!