将PIL numpy 3d数组转换为2d luma值

Question

I've loaded an image using:

import numpy as np
from PIL import Image
imag = Image.open("image.png")    
I = np.asarray(imag)

Where the shape of I is (951, 1200, 3)

But I would like to average each pixel roughly to it's luma values ( (r*g*b)/3 ) to make the shape (951, 1200, 1) .

What is the proper numpy operator to do this?

Answer 1

I think the easiest thing is to use Pillow's built-in conversion to Luminance as follows:

import numpy as np
from PIL import Image

# Load image and convert to luminance, and thence to Numpy array
imag = Image.open("image.png").convert('L')    
I = np.asarray(imag)