Counting values along an axis in a 3D array that are greater than threshold values from a 2D array

Question

I have a 3D array of dimensions (200,200,3). These are images of dimensions (200,200) stacked using numpy.dstack. I would like to count the number of values along axis=2 that are greater than a corresponding 2D threshold array of dimensions (200,200). The output counts array should have dimensions (200,200). Here is my code so far.

import numpy as np

stacked_images=np.random.rand(200,200,3)
threshold=np.random.rand(200,200)
counts=(stacked_images<threshold).sum(axis=2)

I am getting the following error.

ValueError: operands could not be broadcast together with shapes (200,200,3) (200,200)

The code works if threshold is an integer/float value. For example.

threshold=0.3
counts=(stacked_images<threshold).sum(axis=2)

Is there a simple way to do this if threshold is a 2D array? I guess I am not understanding numpy broadcasting rules correctly.

Answer 1

numpy is expecting to make a value by value operation. In your case you seem to be wanting to know if any value in the full Z (axis=2) trace exceeds the equivalent x, y value in threshold.

As so just make sure threshold has the same shape, namely by building a 3D threshold using whatever method you prefer. Since you mentioned numpy.dstack:

import numpy as np

stacked_images = np.random.rand(10, 10, 3)
t = np.random.rand(10, 10)
threshold = np.dstack([t, t, t])
counts = (stacked_images < threshold).sum(axis=2)
print(counts)

, which results in:

[[2 0 3 3 1 3 1 0 1 2]
 [0 1 2 0 0 1 0 0 1 3]
 [2 1 3 0 3 2 1 3 1 3]
 [2 0 0 3 3 2 0 2 0 1]
 [1 3 0 0 0 3 0 2 1 2]
 [1 1 3 2 3 0 0 3 0 3]
 [3 1 0 1 2 0 3 0 0 0]
 [3 1 2 1 3 0 3 2 0 2]
 [3 1 1 2 0 0 1 0 1 0]
 [0 2 2 0 3 0 0 2 3 1]]