Invert binary NumPy array using a mask

Question

I have a binary N-by-N (N~=200) NumPy array populated with zeroes and ones. I would like to apply a Boolean mask and 'swap' the values that correspond to True in the mask, so for example if I had:

arr = np.array([0,0,1,1],
               [0,0,1,1],
               [0,0,1,1],
               [0,0,1,1])
mask = np.array([True,False,True,False],
                [True,False,True,False],
                [True,False,True,False],
                [True,False,True,False],

I would like the resulting array to be:

arr_new = np.array([1,0,0,1],
                   [1,0,0,1],
                   [1,0,0,1],
                   [1,0,0,1])

I started this by initially creating a function swap_cell that swaps between the values, and then followed the approaches in this answer.

def swap_cell(x):
    if x == 1.0:
        return 0.0
    elif x == 0.0:
        return 1.0

arr_new = np.where(mask,swap_cell(arr),arr)

This code returns ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all(), which I understand is because of the 'if' statement in my swap_cell() function. I know there must be a much more pythonic way to accomplish this, but I thought this might work.

Nick Alger · Accepted Answer · 2023-10-20 18:57:52Z

3

If you interpret the first (0,1) array as a boolean array, then you are doing the exclusive or operation of each entry in the first array with each entry of the second array.

Therefore, you can do this with the numpy function np.logical_xor() as follows:

arr = np.array([[0,0,1,1],
                [0,0,1,1],
                [0,0,1,1],
                [0,0,1,1]])
mask = np.array([[True,False,True,False],
                 [True,False,True,False],
                 [True,False,True,False],
                 [True,False,True,False]])

np.array(np.logical_xor(arr, mask), dtype=int)

which returns

array([[1, 0, 0, 1],
       [1, 0, 0, 1],
       [1, 0, 0, 1],
       [1, 0, 0, 1]])

answered Oct 20, 2023 at 18:57

Nick Alger

1,11410 silver badges28 bronze badges

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

1 Comment

mozway Over a year ago

Alternatively: (mask!=arr).astype(int)

pho · Accepted Answer · 2023-10-20 18:49:05Z

A simpler way would be to subtract 1 and take the absolute value:

def swap(x):
    return np.abs(x - 1)

Since this is able to operate on the entire array, you can apply the mask, pass the masked array to swap, and assign the returned value to the masked elements:

arr = np.array([[0,0,1,1],
               [0,0,1,1],
               [0,0,1,1],
               [0,0,1,1]])
mask = np.array([[True,False,True,False],
                [True,False,True,False],
                [True,False,True,False],
                [True,False,True,False]])

arr[mask] = swap(arr[mask])

which results in your expected array:

array([[1, 0, 0, 1],
       [1, 0, 0, 1],
       [1, 0, 0, 1],
       [1, 0, 0, 1]])

Another approach would be to convert the dtype to bool, then invert the array, and finally restore the original dtype:

def swap(x):
    return (~x.astype(bool)).astype(x.dtype)

Collectives™ on Stack Overflow

Invert binary NumPy array using a mask

2 Answers 2

1 Comment

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

1 Comment

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related