CODEWITHSUNDEEP
pythonnumpypython-imaging-library
MeadowMuffins
MeadowMuffins

Reputation: 527

Why different size when converting PIL Image to Numpy array?

I've found it bizarre that numpy arrays and PIL images have different shape, in this case, (H,W) in numpy and (W,H) in PIL. My versions are,

Name: numpy Version: 1.13.3

Name: Pillow Version: 4.1.1

IMG = '/path/to/test-image.jpg'

import numpy as np
from PIL import Image
import matplotlib.pyplot as plt
%matplotlib inline

with Image.open(IMG) as img:
    print img.size
    img_np_f = np.asarray(img, order='F')
    print img_np_f.shape
    img_np_c = np.asarray(img, order='C')
    print img_np_c.shape

    plt.subplot(131)
    plt.imshow(img)
    plt.subplot(132)
    plt.imshow(img_np_f)
    plt.subplot(133)
    plt.imshow(img_np_c)
    plt.show()

The output goes,

(320, 240)
(240, 320)
(240, 320)

However, it seems matplotlib handles it correctly anyway.

enter image description here

Upvotes: 1

Views: 1746

Answers (1)

Ghilas BELHADJ
Ghilas BELHADJ

Reputation: 14096

Because Numpy is not an imaging library.

numpy.ndarray.shape gives the shape in this order (H, W, D) to stay coherent with the terminology used in ndarray's axis (axis=0, axis=1, axis=2)

Upvotes: 1

Related Questions