Image standardisation and normalisation using Numpy and OpenCV

Question

I am trying to standardize and then normalise an image using Numpy and OpenCV in the following manner; however, the image that's output from matplotlib looks identical. Why is that?

Code

%matplotlib inline

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

IMG_SIZE = 256

def show_img(img):
    img_rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
    imgplot = plt.imshow(img_rgb)

img = cv2.imread('/content/drive/My Drive/ai/test_images/test_image3.tif')
img = cv2.resize(img, (IMG_SIZE, IMG_SIZE))
show_img(img)

img = img.astype('float32')
img = (img - img.mean(axis=(0, 1, 2), keepdims=True)) / img.std(axis=(0, 1, 2), keepdims=True)

img = cv2.normalize(img, None, 0, 1, cv2.NORM_MINMAX)
show_img(img)

Simon Delecourt · Accepted Answer · 2019-07-16 09:07:30Z

2

It is a normal behaviour. Normalizing an image means to change its ranges from 0-255 to 0-1. Matplotlib checks the range of the RGB values and display the image accordingly.

As for standardisation, if you look closely you can see a color shift. Orange seems a little lighter on the second image.

These behaviours are normal because preprocessing should not change the image (at least the information inside) but should help the model to processed them ALL.

answered Jul 16, 2019 at 9:07

Simon Delecourt

1,60910 silver badges14 bronze badges

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

Collectives™ on Stack Overflow

Image standardisation and normalisation using Numpy and OpenCV

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related