Matlab / Octave bwdist() in Python or C

While Matlab bwdist returns distances to the closest non-zero cell, Python distance_transform_edt returns distances “to the closest background element”. SciPy documentation is not clear about what it considers to be the “background”, there is some type conversion machinery behind it; in practice 0 is the background, non-zero is the foreground.

So if we have matrix a:

>>> a = np.array(([0,1,0,0,0],
              [1,0,0,0,0],
              [0,0,0,0,1],
              [0,0,0,0,0],
              [0,0,1,0,0]))

then to calculate the same result we need to replaces ones with zeros and zeros with ones, e.g. consider matrix 1-a:

>>> a
array([[0, 1, 0, 0, 0],
       [1, 0, 0, 0, 0],
       [0, 0, 0, 0, 1],
       [0, 0, 0, 0, 0],
       [0, 0, 1, 0, 0]])
>>> 1 - a
array([[1, 0, 1, 1, 1],
       [0, 1, 1, 1, 1],
       [1, 1, 1, 1, 0],
       [1, 1, 1, 1, 1],
       [1, 1, 0, 1, 1]])

In this case scipy.ndimage.morphology.distance_transform_edt gives the expected results:

>>> distance_transform_edt(1-a)
array([[ 1.        ,  0.        ,  1.        ,  2.        ,  2.        ],
       [ 0.        ,  1.        ,  1.41421356,  1.41421356,  1.        ],
       [ 1.        ,  1.41421356,  2.        ,  1.        ,  0.        ],
       [ 2.        ,  1.41421356,  1.        ,  1.41421356,  1.        ],
       [ 2.        ,  1.        ,  0.        ,  1.        ,  2.        ]])

Does scipy.ndimage.morphology.distance_transform_edt meet your needs?

No need to do the 1-a

>>> distance_transform_edt(a==0)
    array([[ 1.        ,  0.        ,  1.        ,  2.        ,  2.        ],
           [ 0.        ,  1.        ,  1.41421356,  1.41421356,  1.        ],
           [ 1.        ,  1.41421356,  2.        ,  1.        ,  0.        ],
           [ 2.        ,  1.41421356,  1.        ,  1.41421356,  1.        ],
           [ 2.        ,  1.        ,  0.        ,  1.        ,  2.        ]])