1 down vote favorite

For example I am trying to find the similarity between two images using skimage - SSIM. The code block will be as follows

from skimage.measure import compare_ssim as ssim
from skimage import io
from skimage.transform import resize

a = io.imread("http://ecx.images-amazon.com/images/I/51PV4Dd8wAL._AC_UL246_SR190,246_.jpg",as_grey=False,dtype="float64")
b = io.imread("http://ecx.images-amazon.com/images/I/914ZHE6JolL._UY500_.jpg",as_grey=False,dtype="float64")

a = resize(a,b.shape)
ssim(a, b,multichannel=True)

I got the similarity score as follows


Although both the images are same where one images in slightly oriented , I am getting very really low scores for this comparisons. Am I missing any preprocessing steps here before I compute compare images. If so what are the things I should consider before comparing images.

Thanks In Advance !!


There is not any fault! This is one of pitfalls of SSIM. It is very sensitive to the geometrical degradations such as orientation!


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.