from scipy.spatial.distance import directed_hausdorff
from .helper import _validate_array_and_make_bool
[docs]def hausdorff_distance(mask_pred, mask_gt):
"""
Calculates the Hausdorff distance for a given image, provided a ground truth. Both arrays must have the same number of columns.
see https://docs.scipy.org/doc/scipy/reference/generated/scipy.spatial.distance.directed_hausdorff.html
Parameters
----------
mask_pred : np.ndarray
predicted segmentation mask
mask_gt : np.ndarray
ground truth mask
Returns
-------
ddouble
The directed Hausdorff distance between arrays u and v,
index_1int
index of point contributing to Hausdorff pair in u
index_2int
index of point contributing to Hausdorff pair in v
"""
mask_pred = _validate_array_and_make_bool(mask_pred)
mask_gt = _validate_array_and_make_bool(mask_gt)
return directed_hausdorff(mask_pred, mask_gt)