Extracting a region from an image using slicing in Python, OpenCV

Question

I have an image and I want to extract a region from it. I have coordinates of left upper corner and right lower corner of this region. In gray scale I do it like this:

I = cv2.imread("lena.png")
I = cv2.cvtColor(I, cv2.COLOR_RGB2GRAY)
region = I[248:280,245:288]
tools.show_1_image_pylab(region)

I can't figure it out how to do it in color. I thought of extracting each channel R, G, B; slicing this region from each of the channels and to merge them back together but there is gotta be a shorter way.

Maybe you could use GetRectSubPix() to get the region instead of slicing. — martineau
– martineau, Commented Feb 25, 2013 at 18:10
I just tested it. And I got color image itself. Display image using cv2.imshow() function, not any pylab or matplotlib function, it is just shown in different color map in them. — Abid Rahman K
– Abid Rahman K, Commented Feb 25, 2013 at 18:49
And if you are doubt that, check region.shape and make sure it is of the format [rows,cols,3]. — Abid Rahman K
– Abid Rahman K, Commented Feb 25, 2013 at 18:50

Abid Rahman K · Accepted Answer · 2014-04-13 10:42:38Z

148

There is a slight difference in pixel ordering in OpenCV and Matplotlib.

OpenCV follows BGR order, while matplotlib likely follows RGB order.

So when you display an image loaded in OpenCV using pylab functions, you may need to convert it into RGB mode. ( I am not sure if any easy method is there). Below method demonstrate it:

import cv2
import numpy as np
import matplotlib.pyplot as plt

img = cv2.imread('messi4.jpg')
b,g,r = cv2.split(img)
img2 = cv2.merge([r,g,b])
plt.subplot(121);plt.imshow(img) # expects distorted color
plt.subplot(122);plt.imshow(img2) # expect true color
plt.show()

cv2.imshow('bgr image',img) # expects true color
cv2.imshow('rgb image',img2) # expects distorted color
cv2.waitKey(0)
cv2.destroyAllWindows()

NB : Please check @Amro 's comment below for better method of conversion between BGR and RGB. img2 = img[:,:,::-1] . Very simple.

Run this code and see the difference in result yourself. Below is what I got :

Using Matplotlib :

enter image description here

Using OpenCV :

enter image description here

edited Apr 13, 2014 at 10:42

answered Feb 25, 2013 at 19:21

Abid Rahman K

52.8k31 gold badges149 silver badges158 bronze badges

Sign up to request clarification or add additional context in comments.

5 Comments

Amro Over a year ago

you could just do: img2 = img[:,:,::-1] to flip the colors dimension from BGR to RGB (using only NumPy indexing)

Abid Rahman K Over a year ago

+1 - yeah, you are right and it would be faster than my answer. Thanks for the comment. (I just wanted to show that flipping of r and b channels specifically.)

vahid abdi Over a year ago

Another way to achieve that result is using cv2.cvtColor with cv2.COLOR_BGR2RGB flag.

Hatshepsut Over a year ago

How about this? cv2.imshow = lambda _, y: plt.imshow(y[:,:,::-1]).figure

bluedroid Over a year ago

it's img2 = img[::,::-1]

noob · Accepted Answer · 2016-10-09 13:05:16Z

50

2 more options not mentioned yet:

img[..., ::-1] # same as the mentioned img[:, :, ::-1] but slightly shorter

and the versatile

cv2.cvtColor(img, cv2.COLOR_BGR2RGB)

edited Oct 9, 2016 at 13:05

noob

3121 gold badge2 silver badges11 bronze badges

answered Mar 1, 2016 at 22:46

Israel Unterman

13.6k4 gold badges30 silver badges35 bronze badges

Comments

Sravan Chittupalli · Accepted Answer · 2020-05-26 17:34:27Z

1

Best way to do this is to use :-
img2 = cv2.cvtColor(img , cv2.COLOR_BGR2RGB)

This will convert the BGR 'img' array to RGB 'img2' array. Now you can use img2 array for imshow() function of matplotlib.

Refer Link:- cvtColor

answered May 26, 2020 at 17:34

Sravan Chittupalli

111 bronze badge

Collectives™ on Stack Overflow

Extracting a region from an image using slicing in Python, OpenCV

3 Answers 3

5 Comments

Comments

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

5 Comments

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related