Image Normalization
Normalizing an Image across Different Axes¶
How does it look when we normalize an image over spatial dimensions instead of color channels?
In [1]:
from IPython.display import display
from tensorflow.keras.preprocessing import image
from tensorflow.keras.utils import normalize
img = image.load_img('dog.jpg')
display(img)
x = image.img_to_array(img)
assert x.shape[-1] == 3, "Not a RGB image with channels last."
# Normalize per row.
display(image.array_to_img(normalize(x, axis=0)))
# Normalize per column.
display(image.array_to_img(normalize(x, axis=1)))
# Normalize per color channel.
display(image.array_to_img(normalize(x, axis=2)))
Comments
Comments powered by Disqus