Edit: I fucked up, the similarity ins in fact 0.7780982999075259 so this means, that 203973 pixels are identical. Originally I forgot to adjust the dimensions in the code.

I had the same thought and did this, I was lazy and just used OpenCV do tho this. The similarity is 0.4024264539258554

The images have a size of 512x512 pixels, so 262144 Pixels in total. So based on the similarity calculated by OpenCV 105493 are identical.

I also have to admit, that doing coding while being stoned is fucking hard.

I used this code:

import cv2

width = 512
height = 512

A  = cv2.imread("Lion.png", cv2.IMREAD_COLOR)
B  = cv2.imread("house.png", cv2.IMREAD_COLOR)


errorL2 = cv2.norm(A, B, cv2.NORM_L2)
similarity = 1 - errorL2 / (height * width)
print("Similarity = ", similarity)