Python interaction between two numpy arrays of the same shape involving cell neighbours

Question

I have two 2d bitewise arrays (1s and 0s only). I am doing

array2 = numpy.where( array1, 0, array2 )

which changes the values of array2 when array1 matches the condition. But what if I don't want to change the values of the very cells where the condition if satisfied, but their neighbors cells? I mean, instead of i,j for which array1 evals to True, I want to change the subarray array2[i-1:i+2,j-1:j+2]. It is possible to avoid looping, which is really slow on big arrays?

Answer 1

see scipy.ndimage.filters.generic_filter

so in your example, the footprint argument would be (3, 3), that is a neighborhood of 3 by 3 and the the function argument should return if the condition is true or false;

something like:

func = lambda xs: np.any(xs)  # or whatever appropriate
mask = generic_filter(array1, func, footprint=(3, 3))
array2[mask] = 0