################################################
SOMart (Self Organizing Art) Alpha Version 
################################################

SOMart was developped with the goal of automatically cluster in different groups art
images which were made using different techniques. The algorithm had
good results in tasks like distinguishing wood engraves (xilographs)
from litographs.

This project is based on Python ia636 (Image Processing module) and
SOM_PAK 3.2 (Self Organizing Maps toolkit).

It works extracting texture features from JPEG images using
Co-occurence Matrixes and Fourier Analysis (FFT) to feed a Self
Organizing Map (non supervised neural network) to automatically
classificate the images based on input extraction vectors.

#################################################
Dependencies
#################################################

- Netpbm - http://netpbm.sourceforge.net/
- Imagemagick - http://www.imagemagick.org/script/binary-releases.php
- Python HTML module - http://pypi.python.org/pypi/html#downloads
- IA636 module - http://www.dca.fee.unicamp.br/ia636/README.html

* SOM_pak 3.2 is distributed together with the project

#################################################
Configuration
#################################################

As this is still an aplha version, the configuration of self organizing
maps and feature extraction parameters can be made only by editing main.py file and
changing values on somart class constructor variables.

You can choose an image set changing the value of somart.testimages
("escher_engraves" by default).

If you wanna make tests with your own images, you have to create a new
directory inside imgs with this structure:

imgs/$image_set_name
imgs/$image_set_name/jpg  <-- put your JPEG files here
imgs/$image_set_name/pgm  <-- leave this empty, PGM files will be stored here

You will also need to create this directory:

imgs/$image_set_name

And then you will need to change the value of somart.testimages to the
name of the created directory ($image_set_name).

* Automatic creation of necessary directories will be working on
  beta version.

#################################################
Running
#################################################

To run the main algorithm:

$ cd main
$ python main.py

After running main.py, the result of image clustering can be viewed
in a HTML file named data/$set/$file.html, where $set is the
name given to the image test set and $file.html is the filename
generated according to the current date/time when main.py was called. 

* Currently SOMart only runs in POSIX-like systems because it uses
  shell scripting routines.