porespy.filters.local_thickness#

porespy.filters.local_thickness(im: numpy.typing.NDArray, dt: numpy.typing.NDArray = None, method: Literal['bf', 'imj', 'conv', 'dt'] = 'dt', smooth: bool = True, mask: numpy.typing.NDArray = None, approx: bool = False, sizes: int = 25)#

Insert a maximally inscribed sphere at every pixel labelled by sphere radius

This is a wrapper method for computing local thickness via a variety of different methods.

Parameters:

  • im (ndarray) – Boolean image of the porous material

  • dt (ndarray, optional) – The distance transform of the image

  • method (str) –

    Which method to use to compute the local thickness. Options are:

    Method

    Description

    ’dt’

    Uses distance transforms to perform erosion and dilation for each radius in the image

    ’bf’

    Uses brute-force to inserts spheres at each voxel

    ’imj’

    Uses the brute-force method but reduces the number of insertion sites by 80-90% to speed up the process

    ’conv’

    Uses FFT-based convolution to perform erosion and dilation for each radius in the image

  • sizes (array_like or scalar) – This is only used if the method is dt or conv. If a list of values is provided they are used directly. If a scalar is provided then that number of points spanning the min and max of the distance transform are used. If None, then all the unique values in the distance transform are used, which may become time consuming. This can be sped up if dt is provided and rounded to the nearest integer first.

  • smooth (bool, optional) – Indicates if protrusions should be removed from the faces of the spheres or not. Default is True.

  • mask (ndarray, optional) – This is only used if the method is bf or imj. A boolean mask indicating which sites to insert spheres at. If not provided then all True values in im are used.

  • approx (bool, optional) – This is only used if the method is imj. If True the algorithm is more agressive at skipping voxels to process, which speeds things up, but this sacrifices accuracy in terms of a voxel-by-voxel match with the reference implementation. The default is False, meaning full accuracy is the default.

Returns:

lt – The local thickness of the image with each voxel labelled according to the radius of the largest sphere which overlaps it

Return type:

ndarray

On this page