Image does not load as grayscale (skimage)

rdv picture rdv · Jun 15, 2017 · Viewed 10.1k times · Source

I'm trying to load an image as grayscale as follows:

from skimage import data
from skimage.viewer import ImageViewer

img = data.imread('my_image.png', as_gray=True)

However, if I check for its shape using img.shape it turns out to be a three-dimensional, and not two-dimensional, array. What am I doing wrong?

Answer

Tonechas picture Tonechas · Jun 19, 2017

From scikit-image documentation, the signature of data.imread is as follows:

skimage.data.imread(fname, as_grey=False, plugin=None, flatten=None, **plugin_args)

Your code does not work properly because the keyword argument as_grey is misspelled (you put as_gray).

Sample run

In [4]: from skimage import data

In [5]: img_3d = data.imread('my_image.png', as_grey=False)

In [6]: img_3d.dtype
Out[6]: dtype('uint8')

In [7]: img_3d.shape
Out[7]: (256L, 640L, 3L)

In [8]: img_2d = data.imread('my_image.png', as_grey=True)

In [9]: img_2d.dtype
Out[9]: dtype('float64')

In [10]: img_2d.shape
Out[10]: (256L, 640L)