Image Similarity or Distance

	from scipy import ndimage import numpy as np
	from scipy import misc
	import PIL
	import matplotlib.pyplot as plt
	import os
	import numpy as np
	from PIL import Image
	from io import StringIO
	import urllib
	os.chdir('_______')
	img = Image.open("pool1.png" )
	img.size
	data = np.asarray( img, dtype='uint8' )
	data.shape
	data[:,:,3].shape
	PIL.Image.fromarray(data[:,:,3])
	PIL.Image.fromarray(data[:,:,2])
	PIL.Image.fromarray(data[:,:,1])
	def get_thumbnail(image, size=(128,128), stretch_to_fit=False, greyscale=False):
	" get a smaller version of the image - makes comparison much faster/easier"
	if not stretch_to_fit:
	image.thumbnail(size, Image.ANTIALIAS)
	else:
	image = image.resize(size); # for faster computation
	if greyscale:
	image = image.convert("L") # Convert it to grayscale.
	return image
	img1 = get_thumbnail(Image.open('pool1.png'))
	img2 = get_thumbnail(Image.open('pool2.png'))
	np.asarray(img1, dtype='uint8' ).shape
	np.asarray(img2, dtype='uint8' ).shape
	def mse(imageA, imageB):
	# the 'Mean Squared Error' between the two images is the
	# sum of the squared difference between the two images;
	# NOTE: the two images must have the same dimension
	imageA_np = np.asarray(imageA, dtype='uint8' )[:,:,0:3]
	imageB_np = np.asarray(imageB, dtype='uint8' )[:,:,0:3]
	err = np.sum((imageA_np.astype("float") - imageB_np.astype("float")) ** 2)
	err /= float(imageA_np.shape[0] * imageA_np.shape[1])
	# return the MSE, the lower the error, the more "similar"
	# the two images are
	return err
	mse(img1, img2)
	# normalized cross correlation
	temp = []
	def get_pixel_avg_vec(imageobj):
	from numpy import average
	vector = []
	for pixel_tuple in imageobj.getdata():
	vector.append(average(pixel_tuple))
	return vector
	get_pixel_avg_vec(pool_az)
	def cross_corr(imageA, imageB):
	from numpy import linalg, dot
	# Get average vector of image
	vec_A = get_pixel_avg_vec(imageA)
	vec_B = get_pixel_avg_vec(imageB)
	# Get L2 norm
	a_norm = linalg.norm(vec_A, 2)
	b_norm = linalg.norm(vec_B, 2)
	# Cosine Similarity
	sim = dot(vec_A / a_norm, vec_B / b_norm)
	return sim
	cross_corr(pool_az, pool_ggl1), cross_corr(pool_az, pool_ggl2)
	mse(pool_az, pool_ggl1), mse(pool_az, pool_ggl2)
	from scipy.stsci.convolve import cross_correlate
	cross_correlate(img_az, mode=FULL)