I got a result of 0.9871965646743774....

[c-yn]

Hello, I ran your code and printed a result of 0.9871965646743774...

[mayorx]

Hi, morgen-star,

This huge gap may be due to the compression/change of the picture?
After reading these images, try this:

import hashlib

print(hashlib.md5(inp).hexdigest())
print(hashlib.md5(tgt).hexdigest())

and there is my output:

802d2bbf31c45c0f9fc5c3d3457e48f1
a94cbc92fd389b10f33757f55d0a6ee8

[c-yn]

12e93f9a8650872d55755d6e3ac8bfc3
fced9ad1fd7ebe101217cac30bbd20a3

Thanks for your reply. Above is my output.
As you said, maybe the performance gap is due to the change of downloaded pictures.
I will test on other data and compare the outcome with matlab version.
Anyway, thanks for your code.