如何更改只有 0 和 1 的 2d numpy 数组中所有 1 的索引？

Question

I have a 2d numpy array(arr) which only has 0s and 1s.

For example, a 2d numpy array in shape(h,w).

I want to resize the array to shape(h // scale, w // scale), and I need to keep all the 1s.

# arr is a 2d numpy array
h, w = arr.shape
h_new, w_new = h // scale, w // scale
arr_new = np.zeros((h_new, w_new))
for i in range(h):
    for j in range(w):
        if arr[i, j] == 1:
            arr_new[i // scale, j // scale] = 1

For example, a 2d array like this, and scale=2:

[
    [0,0,0,0],
    [0,1,0,0],
    [1,0,0,1],
    [0,0,0,1]
]

After runnung the code, arr_new will be:

[
    [1,0],
    [1,1]
]

the change of all the 1s' coordinates:

(1,1) -> (0,0)
(2,0) -> (1,0)
(2,3) -> (1,1)
(3,3) -> (1,1)

Is there a better way to do this?

Thanks.

Answer 1

Just reshape the array by splitting each dimension to grid (SIZE // SCALE, SCALE) . Next reduce all dimensions of size SCALE using max() to let 1 dominate the SCALExSCALE cell.

arr.reshape(h//scale, scale, w // scale, scale).max(axis=(1,3))

Note that this solution requires shape of array to be multiplicity os (scale, scale) .