create an animation showing Discrete Cosine Transform and what its frequency levels look like

dct-demo.py

#!/usr/bin/python

import sys, os
from PIL import Image
import numpy
import scipy.fftpack

sourceImage = sys.argv[1]

image = Image.open(sourceImage)
image = image.resize( (128,128), 1 )
image = image.convert("L")

dctSize = image.size[0]

# get raw pixel values:
pixels = numpy.array(image.getdata(), dtype=numpy.float).reshape((dctSize, dctSize))

# perform 2-dimensional DCT (discrete cosine transform):
dct = scipy.fftpack.dct(scipy.fftpack.dct(pixels.T, norm="ortho").T, norm="ortho")

# create a series of images with increasingly larger parts of the DCT values being used:
os.mkdir("frames/")
for i in range(0, dctSize):
    dct2 = dct.copy()

    # zero out part of the higher frequencies of the DCT values:
    dct2[i:,:] = 0
    dct2[:,i:] = 0

    # perform 2d inverse DCT to get pixels back:
    idct = scipy.fftpack.idct(scipy.fftpack.idct(dct2.T, norm='ortho').T, norm='ortho')

    # clip/convert pixel values obtained by IDCT, and create image:
    idct = idct.clip(0, 255)
    idct = idct.astype("uint8")
    img = Image.fromarray(idct)

    print img
    img.save("frames/img_%04d.png" % i)


os.system("convert -delay 30 -comment 'example of Discrete Cosine Transform (source image: %s)' frames/img_*.png dct.gif" % sourceImage)

http://imgur.com/mt9xSa9 shows the results for http://commons.wikimedia.org/wiki/File:Lichtenstein_img_processing_test.png