How to get 2D histogram map of an image?

Question

To be more specific, here is the exact requirement. I'm not sure how to word the question. I have an image, of size say (500,500) . I extract only r and g channels

r = image[:, :, 0]
g = image[:, :, 1]

Then, I compute the 2D histogram of r and g

hist2d = np.histogram2d(r, g, bins=256, range=[(255,255),(255,255)])

Now, hist2d[0].shape is (256, 256) since It corresponds to every pair of 256x256 colors. Fine

The main requirement is, in an separate image, called result with same shape as original image ie (500, 500) , I want to populate each element of result with the value of 2d histogram of r and g channels

For example, if r[200,200] is 23 and g[200, 200] is 26, I want to place result[200, 200] = hist2d[0][23, 26]

The naive method for doing this is, simple python loop.

for i in range(r.shape[0]):
    for j in range(r.shape[1]):
        result[i, j] = hist2d[0][r[i, j], g[i, j]]

But for a large image, this takes a significant time to compute. Is there a numpy way of doing this?

Thanks

Answer 1

just use hist2d[0][r, g] :

import numpy as np

r, g, b = np.random.randint(0, 256, size=(3, 500, 500)).astype(np.uint8)
hist2d = np.histogram2d(r.ravel(), g.ravel(), bins=256, range=[[0, 256], [0, 256]])
hist2d[0][r, g]