In [22]:

    

%matplotlib inline
import numpy as np

import matplotlib.pyplot as plt
from PIL import Image, ImageOps

from ssim import SSIM
from ssim.utils import get_gaussian_kernel

gaussian_kernel_sigma = 1.5
gaussian_kernel_width = 11
gaussian_kernel_1d = get_gaussian_kernel(gaussian_kernel_width, gaussian_kernel_sigma)


    

In [23]:

    

size = (256,256)

im = Image.open('test-images/test3-orig.jpg')
im = im.resize(size, Image.ANTIALIAS)

# slightly rotated image
im_rot = Image.open('test-images/test3-rot.jpg')
im_rot = im_rot.resize(size, Image.ANTIALIAS)

# slightly modified lighting conditions
im_lig = Image.open('test-images/test3-lig.jpg')
im_lig = im_lig.resize(size, Image.ANTIALIAS)

# image cropped
im_cro = Image.open('test-images/test3-cro.jpg')
im_cro = im_cro.resize(size, Image.ANTIALIAS)


    

In [24]:

    

im_gray = ImageOps.grayscale(im)
plt.imshow(np.asarray(im_gray),cmap='gray',vmin=0,vmax=255)


    

    
Out[24]:





<matplotlib.image.AxesImage at 0x7f2a11963ba8>






    

In [25]:

    

im_rot_gray = ImageOps.grayscale(im_rot)
plt.imshow(np.asarray(im_rot_gray),cmap='gray',vmin=0,vmax=255)


    

    
Out[25]:





<matplotlib.image.AxesImage at 0x7f2a11a13080>






    

In [26]:

    

im_lig_gray = ImageOps.grayscale(im_lig)
plt.imshow(np.asarray(im_lig_gray),cmap='gray',vmin=0,vmax=255)


    

    
Out[26]:





<matplotlib.image.AxesImage at 0x7f2a11a8ae80>






    

In [27]:

    

im_cro_gray = ImageOps.grayscale(im_cro)
plt.imshow(np.asarray(im_cro_gray),cmap='gray',vmin=0,vmax=255)


    

    
Out[27]:





<matplotlib.image.AxesImage at 0x7f2a11c10470>






    

In [28]:

    

ssim_rot = SSIM(im, gaussian_kernel_1d).ssim_value(im_rot)
ssim_lig = SSIM(im, gaussian_kernel_1d).ssim_value(im_lig)
ssim_cro = SSIM(im, gaussian_kernel_1d).ssim_value(im_cro)

print("SSIM of rotated image %.4f" % ssim_rot)
print("SSIM of modified lighting image %.4f" % ssim_lig)
print("SSIM of cropped image %.4f" % ssim_cro)


    

    




SSIM of rotated image 0.4971
SSIM of modified lighting image 0.8603
SSIM of cropped image 0.3958

In [29]:

    

cw_ssim_rot = SSIM(im).cw_ssim_value(im_rot)
cw_ssim_lig = SSIM(im).cw_ssim_value(im_lig)
cw_ssim_cro = SSIM(im).cw_ssim_value(im_cro)

print("CW-SSIM of rotated image %.4f" % cw_ssim_rot)
print("CW-SSIM of modified lighting image %.4f" % cw_ssim_lig)
print("CW-SSIM of cropped image %.4f" % cw_ssim_cro)


    

    




CW-SSIM of rotated image 0.8724
CW-SSIM of modified lighting image 0.9668
CW-SSIM of cropped image 0.7171