CS376A Digital Image Processing
Skidmore College

Assignment 03
Due: 
Part 1 due April 12 by 11:59 p.m. to meckmann@skidmore.edu

=============

PART 1

1. Finish implementation of K-means clustering
   - to visualize your clusters, assign each class a different color
     and output an image
   - if use R, G and B only as your clustering features, then
     the average (mean) color of each cluster is good choice for 
     display
   - if use any other features, visualize using some premade
     standard set of colors (have 16 prepared) and if say your K
     is 5, only use the first 5,  if K is 16 use all 16 colors

   - implement k-means clustering on R, G, and B as the three 
     features
        - run it on several images with different values of k
          e.g. k=3,4,5,6 for 3 of your own images
     - if image file name was image.jpg, name your output images
        color-3means-image.jpg
        color-4means-image.jpg
        color-5means-image.jpg
        color-6means-image.jpg

   - also do 2 features: intensity and standard deviation of intensity
     in a 5x5 neighborhood
        - run it on several images with different values of k
          e.g. k=3,4,5,6 for 3 of your own images
     - if image file name was image.jpg, name your output images
        intensity-and-texture-3means-image.jpg
        intensity-and-texture-4means-image.jpg
        intensity-and-texture-5means-image.jpg
        intensity-and-texture-6means-image.jpg

     Note well: make the intensity range from 0..1 (scale by dividing calculated
        intensity by 255 --- make sure you use double division (cast on to a double)) 
        BUT you should also make the standard deviation be
        in the range 0..1 --- to do this you'll need to figure out the min
        and max standard deviations and then scale your range of standard 
        deviations to the range 0..1

        e.g. if you find the minimum standard deviation to be minSD and the 
             maximum standard deviation to be maxSD, then to get any of the
             SDs in the 0..1 range, do:  (SD-min)/(max-min)

   - also do 1 feature: standard deviation of intensity in a 5x5 
     neighborhood  
        - run it on several images with different values of k
          e.g. k=3,4,5,6 for 3 of your own images
     - if image file name was image.jpg, name your output images
        texture-3means-image.jpg
        texture-4means-image.jpg
        texture-5means-image.jpg
        texture-6means-image.jpg

   - have a string as the third command line parameter with the following
     possible values:
      "color"
      "intensity-and-texture"
      "texture"

     use the value of that third command line parameter to determine which
     of the three different sets of features to use to cluster

===========

Attach a zip file (please do not create a rar file) with:

  all the java files necessary for me to compile your program

  and some original images and the appropriately named resulting clustered images