numpy: unpack array along an axis

Question

Say that I have a RGB image:

from skimage import data
img = data.astronaut()
print(img.shape)  # (512, 512, 3)

Is there a succinct numpy command to unpack it along the color channels:

R, G, B = np.unpack(img, 2)  # ?

What I am doing is using comprehension:

R, G, B = (img[:, :, i] for i in range(3))

But is there no simpler command?

Divakar · Accepted Answer · 2016-09-20 15:06:21Z

4

Alternatively you can use np.rollaxis -

R,G,B = np.rollaxis(img,2)

answered Sep 20, 2016 at 15:06

Divakar

222k19 gold badges273 silver badges374 bronze badges

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

Comments

David Z · Accepted Answer · 2016-09-20 14:43:40Z

1

You can transpose the length-3 dimension to the front and then unpack it:

R, G, B = img.transpose((2, 0, 1))

answered Sep 20, 2016 at 14:43

David Z

133k29 gold badges264 silver badges284 bronze badges

I will give some time for anyone else to throw a suggestion, but that is pretty clever. :)

Conchylicultor · Accepted Answer · 2017-03-02 18:47:39Z

1

Alternatively, you can use np.split:

R, G, B = np.split(img, img.shape[-1], axis=-1)

If your array is of shape (height, width, channel), you can use np.dsplit to split along the depth dimension:

R, G, B = np.dsplit(img, img.shape[-1])

answered Dec 4, 2016 at 21:10

Conchylicultor

5,9893 gold badges47 silver badges48 bronze badges