### Instagram Unshredder Solution
see http://instagram-engineering.tumblr.com/post/12651721845/instagram-engineering-challenge-the-unshredder

author: Max Brauer
requires: PIL and NumPy

The basic assumption is that when looking at pixel values they will unlikely
change rapidly locally. Thats means that taking two neighboring pixels they
values will be relatively close concerning euclidean distance in RGBA space.
Even though it is safer to say that is very unlikely that the will be far apart.
If so it would be evidence for an edge within the image. When looking at the
scrambled columns it is very likely that those columns belong together which
most outward pixels columns are closest together regarding euclidean distance.
Thereby knowing the column width a picture might be unshreddered.

The approach to column width detection does not always work. If it's not working
reshredder the original image and try again.