How to get color of all pixels in an area in python

Question

I'm trying to write a program on linux that does something if the pixels in an area aren't all the same color, for example:

if color not "255, 255, 255":
    #do something

this is what i have for one pixel:

import time, pyautogui
time.clock()
image = pyautogui.screenshot()
color = image.getpixel((1006, 553))
print(time.clock())
print(color)

I know how to get the color of a pixel using .getpixel() but that only gets one pixel

Basically, how do i get the color of an area of pixels when i know all the pixels in that area are the same color.

Also, as quick as possible, like 0.5s or under.

Consider using a ranged `for` loop to go over all pixels in the area - if one or more doesn't match the previous pixel, the check fails. — Nick Reed, Oct 16 '19 at 19:00
Depends how big your image is. 0.5 seconds is a LOT of time for a program - it should be enough for any image of realistic size. Are you *certain* the area is all the same pixel, though, or is that what your code is trying to determine? — Nick Reed, Oct 16 '19 at 19:06
What libraries are you using? I'm sure there are many different ones who have a `.getpixel()` method. A runtime goal of 0.5s is meaningless if we don't know anything about your data! — AMC, Oct 16 '19 at 19:06
@Fízli So like 100k pixels. On my phone, a comparable operation takes 0.0112 seconds to process the data (timed using `timeit`, `n=1000`) That doesn't include the time needed to get the data though since that's highly dependant. — Carcigenicate, Oct 16 '19 at 19:20
Just use Numpy (`np.var`) to calculate the variance of any subimage. If it is zero, there is no variation in the colour. — Mark Setchell, Oct 17 '19 at 18:59

score 0 · Answer 1 · answered Oct 17 '19 at 18:14

I keep recommending it, but the scikit-image library is pretty great, and they have some really solid documentation and examples. I would recommend a combo of that and using numpy arrays directly. It is just a lot faster when working directly with pixels. You will have to convert the PIL image to a numpy array...but this should work with that:

import pyautogui
import numpy as np

image = pyautogui.screenshot()
np_image = np.array(image)

You can slice the image:

red_slice = np_image[0:50, 0:50,0]
red_mask = red_slice == 200

This would give you the values for red in the upper right 50x50 pixel area. red_mask is an array of True/False values whether each red value in that area is equal to 200. This can be repeated for the other channels as you see fit.

How to get color of all pixels in an area in python

1 Answers1