How to convert PIL image to numpy array? [duplicate]

Question

When I do np.asarray(my_img) or array(my_img).shape it returns (2412L, 3600L, 3L) on a jpg image,but I just want a 2D (2412L, 3600L) array, so how to convert it correctly? Thanks in advance.

my_image = "AI.jpg"

from matplotlib.pyplot import imread
from PIL import Image

fname = "images/" + my_image
image = Image.open(fname) 
print(image.size)    # output: (3600, 2412)
print(np.asarray(image).shape) # output: (2412L, 3600L, 3L)
print(np.array(image).shape) # output: (2412L, 3600L, 3L)

Presumably what you have is RGB image, hence the 3 colour channels (that's what the 3rd dimension in that array represents). — Dan Mašek
– Dan Mašek, Commented Feb 10, 2020 at 14:05
image.size shows only (x,y) but numpy's shape shows (x,y,channels) - so it is correct result. shape can show (x,y) only if you have single value per pixel - it means greyscale — furas
– furas, Commented Feb 10, 2020 at 14:08
but I know "my_image = scipy.misc.imresize(image, size=(num_px,num_px)).reshape((1, num_px*num_px*3)).T“ can convert it to 2D ( but imresize() deprecated), so how can i do that? — harrison dong
– harrison dong, Commented Feb 10, 2020 at 14:27
Does this answer your question? How to convert a PIL Image into a numpy array? — Alperen Kantarcı
– Alperen Kantarcı, Commented Feb 10, 2020 at 15:21

Mark Setchell · Accepted Answer · 2020-02-10 14:28:31Z

1

If your image dimensions have a 3 for the final axis, that normally means you have a 3-channel RGB image.

If you want a single channel image, you will lose the RGB colour and just have a greyscale image. You can do that like this:

grey = Image.open(fname).convert('L')

answered Feb 10, 2020 at 14:28

Mark Setchell

210k32 gold badges310 silver badges504 bronze badges

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

3 Comments

harrison dong Over a year ago

so if use deprecated imresize() (my_image = scipy.misc.imresize(image, size=(num_px,num_px)).reshape((1, num_px*num_px*3)).T), it also lose the RGB color and have greyscale image?

Mark Setchell Over a year ago

I don't understand what the connection is to imresize(). Your question is about going from 3 channels to 1 channel. That is normally a greyscale conversion, or the extraction or calculation of a new channel. imresize() is for changing the height and/or the width of the image.

harrison dong Over a year ago

It's a duplicated question. so pls visit: stackoverflow.com/questions/384759/… to get the answer.

Collectives™ on Stack Overflow

How to convert PIL image to numpy array? [duplicate]

1 Answer 1

3 Comments

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

3 Comments

Linked

Related