Replace each pixel in a numpy array with a formula

Question

I have a numpy array, which represent image pixels, shape of the array is (500, 800, 3) I need to multiply each pixel with the formula:

pixel_color = r * 0.299 + g * 0.587 + b * 0.114

So, [r, g, b] -> [pixel_color, pixel_color, pixel_color]

How can I do this with numpy operations?

Does the last (inner) dimension contain the rgb values? And if so, do you want to end up with a two dimensional array? — 9769953
– 9769953, Commented Jul 23, 2018 at 13:42
There are no grayscale shades of rgb; that doesn't mean anything. Do you mean the (relative) values of the three colours, or a three dimensional grayscale picture? — 9769953
– 9769953, Commented Jul 23, 2018 at 13:44

swhat · Accepted Answer · 2018-07-23 14:24:46Z

1

You're looking for the weighted average of pixels.

a = np.ones(shape=(500,800,3))
gray_image = np.average(a, axis=2, weights=[.299, .587, .114])

Then, to get back to the original shape, you can use np.repeat with a new axis.

np.repeat(gray_image[:,:,np.newaxis], repeats=3, axis=2)

answered Jul 23, 2018 at 14:09

swhat

4933 silver badges12 bronze badges

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

1 Comment

Thanks for the np.repeat example, very relevant here!