Same image captured in different time isn't matching in pillow

Question

I have a program where I have to exit the program if image1 and image2 are the same. So, I took two same screenshot but in different time, but they aren't matching. Here is my code:

import pyautogui
import time
from PIL import Image

time.sleep(3)




image1 = Image.open('my_screenshot0.png')
image2 = Image.open('my_screenshot1.png')


if image1 == image2:
    print("matched")
else:
    print("Failed")

what is the problem and how to solve it ?

Edit:

I ran this code and the result is still "False"..

import pyautogui
import numpy as np


pyautogui.screenshot('test1.png')
pyautogui.screenshot('test2.png')

if np.array('test1.png') == np.array('test2.png'):
    print('TRUE')
else:
    print("False")

Images are included:

This has been already answered in https://stackoverflow.com/questions/35176639/compare-images-python-pil, check it and come back if its not working — Ghost Ops, Aug 19 '21 at 13:21
How can you be sure that the images are exactly equal? To be sure try comparing the pixel values as lumpy arrays `np.array(image1) == np.array(image2)` — Matteo Zanoni, Aug 19 '21 at 13:24
@MatteoZanoni i am sure because i used pyautogui to take the screenshots and compare at the same time — Asir Shahriar Roudra, Aug 19 '21 at 13:35
Hopefully there was no moving video, no clock and no mouse in your images... share the images and we will be able to tell you the difference. — Mark Setchell, Aug 19 '21 at 13:38
Maybe the problem is that the object must be the exact same, because both these images take up different areas in memory, so I think that's the problem. — LuckElixir, Aug 19 '21 at 13:57
The difference is in the *"cloud sync"* icon immediately to the left of the clock at bottom right. You can find it immediately with **ImageMagick** in Terminal with `magick a.png b.png -compose difference -composite result.png` — Mark Setchell, Aug 19 '21 at 13:58
@MarkSetchell I resolved the cloud sync , but the problem again appears , check the result and test1, test2 pictures. link: https://imgur.com/a/Fm9JOyx — Asir Shahriar Roudra, Aug 19 '21 at 14:24
There's a difference in the vertical bar down the right side of the images, try emphasising the difference with `magick hkdD8Qe.png SK9XVz9.png -compose difference -composite -normalize result.png` — Mark Setchell, Aug 19 '21 at 14:28
Does this answer your question? [Compare images Python PIL](https://stackoverflow.com/questions/35176639/compare-images-python-pil) — pippo1980, Aug 19 '21 at 23:46

pippo1980 · Accepted Answer · 2021-08-20T15:12:46.280

ok re-tried both approaches, here code and output:


from PIL import Image

import numpy as np


image1 = Image.open('test1.png')
image1c = Image.open('test1c.png')
image2 = Image.open('test2.png')
image3 = Image.open('test3.png')

if image1 == image1c:
    print("matched")
else:
    print("Failed")

if image1 == image2:
    print("matched")
else:
    print("Failed")
    
if image1 == image3:
    print("matched")
else:
    print("Failed")
    
print('_________________________________')


img1 = np.array(image1)
img1c = np.array(image1c)
img2 = np.array(image2)
img3 = np.array(image3)

if np.array_equal(img1,img1c) : 
    print('TRUE')
else:
    print("False")
    
if np.array_equal(img1,img2) : 
    print('TRUE')
else:
    print("False")
    
if np.array_equal(img1,img3) : 
    print('TRUE')
else:
    print("False")

output:

matched
Failed
Failed
_________________________________
TRUE
False
False

so seems problem where in images, You can compare pillow Images and if converted to numpy arrays need to use

numpy.array_equal

to compare them

here the images used (test1.png is your first image, test1c.png its a copy of it, test2.png is your second image, test3.png its a modified by me image):

test1.png:

test1c.png:

test2.png:

test3.png:

let me know if something is wrong with it

took a little bit , but found different pixels in images:

here the pics comparison:

my eyes are burning from eye strain ;-)

Same image captured in different time isn't matching in pillow

1 Answers1