How do I make rectangular image squared using OpenCV and Python?

Question

I have all sorts of images of rectangular shape. I need to modify them to uniform square shape (different size ok).

For that I have to layer it on top of larger squared shape. Background is black.

I figured it to the point when I need to layer 2 images:

import cv2
import numpy as np
if 1:
        img = cv2.imread(in_img)
        #get size
        height, width, channels = img.shape
        print (in_img,height, width, channels)
        # Create a black image
        x = height if height > width else width
        y = height if height > width else width
        square= np.zeros((x,y,3), np.uint8)
        cv2.imshow("original", img)
        cv2.imshow("black square", square)
        cv2.waitKey(0)

How do I stack them on top of each other so original image is centered vertically and horizontally on top of black shape?

Possible duplicate of [resize image canvas to maintain square aspect ratio in Python, OpenCv](https://stackoverflow.com/questions/44720580/resize-image-canvas-to-maintain-square-aspect-ratio-in-python-opencv) — alkasm, Aug 12 '17 at 07:22

score 2 · Accepted Answer · edited Jan 22 '20 at 09:30

2

I figured it. You need to "broadcast into shape":

square[(y-height)/2:y-(y-height)/2, (x-width)/2:x-(x-width)/2] = img

Final draft:

import cv2
import numpy as np
if 1:
        img = cv2.imread(in_img)
        #get size
        height, width, channels = img.shape
        print (in_img,height, width, channels)
        # Create a black image
        x = height if height > width else width
        y = height if height > width else width
        square= np.zeros((x,y,3), np.uint8)
        #
        #This does the job
        #
        square[int((y-height)/2):int(y-(y-height)/2), int((x-width)/2):int(x-(x-width)/2)] = img
        cv2.imwrite(out_img,square)
        cv2.imshow("original", img)
        cv2.imshow("black square", square)
        cv2.waitKey(0)

edited Jan 22 '20 at 09:30

Parikshit Chalke

3,745
2
16
26

answered Aug 12 '17 at 03:41

Alex B

2,165
2
27
37

1

How to create white background? OR simply background of any other color? – Parikshit Chalke Jan 22 '20 at 06:31
@ParikshitChalke To get any color, replace `square= np.zeros((x,y,3), np.uint8)` with: `square= np.ones((x,y,3), np.uint8) * (B, G, R)` Where B, G, R are the standard RGB values for colors with integers between 0 and 255. For Grayscale images replace these lines: `height, width = img.shape` `print (in_img,height, width)` `square = np.ones((x,y), np.uint8) * value` Where 'value' is the greyscale value for the background (0 for black and 255 for white) – Snery May 24 '23 at 12:13

score -5 · Answer 2 · answered Aug 15 '17 at 06:26

-5

You can use numpy.vstack to stack your images vertically and numpy.hstack to stack your images horizontally.

Please mark answered if you this resolves your problem.

answered Aug 15 '17 at 06:26

Muhammad Abdullah

876
2
8
26

How do I make rectangular image squared using OpenCV and Python?

2 Answers2