Overview
--------

affine_registration is a quick-and-dirty set of scripts for automatically transforming a set of images such that a selected set of points is aligned as best as possible in the output.  

Example Use Case
----------------

Say you have a bunch of pictures of a window that were all taken from slightly different angles. 

- Run `./label.py /path/to/image/directory` and manually mark the four corners of the window frame in each picture (ensuring that you follow the same order each time).

- Run `./transform.py /path/to/image/directory` to generate the transformed output images.  Each individual image will be transformed so as to try to have its registration points overlap with the corresponding registration points in the first image.
 
NOTE: An affine transform is probably not sufficient to make all your images perfect, as it does not take into account camera optics or anything like that.  It was simply chosen as a simple, good-enough solution for many cases.

Dependencies
------------
- numpy
- scipy
- matplotlib
- Python Imaging Library (PIL)