def __init__(
self,
num_control_points: Union[int, Tuple[int, int, int]] = 7,
max_displacement: Union[float, Tuple[float, float, float]] = 7.5,
locked_borders: int = 2,
image_interpolation: Interpolation = Interpolation.LINEAR,
p: float = 1,
seed: Optional[int] = None,
is_tensor = False,
):
super().__init__(p=p, seed=seed, is_tensor=is_tensor)
self._bspline_transformation = None
self.num_control_points = to_tuple(num_control_points, length=3)
self.parse_control_points(self.num_control_points)
self.max_displacement = to_tuple(max_displacement, length=3)
self.parse_max_displacement(self.max_displacement)
self.num_locked_borders = locked_borders
self.is_tensor = is_tensor
if locked_borders not in (0, 1, 2):
raise ValueError('locked_borders must be 0, 1, or 2')
if locked_borders == 2 and 4 in self.num_control_points:
message = (
'Setting locked_borders to 2 and using less than 5 control'
'points results in an identity transform. Lock fewer borders'
' or use more control points.'
)
raise ValueError(message)
self.interpolation = self.parse_interpolation(image_interpolation)
def __init__(
self,
data: TypeData,
patch_size: TypeTuple,
patch_overlap: TypeTuple,
):
self.array = data
patch_size = to_tuple(patch_size, n=3)
patch_overlap = to_tuple(patch_overlap, n=3)
self.locations = self._grid_spatial_coordinates(
self.array,
patch_size,
patch_overlap,
)
def __init__(
self,
data: Union[np.ndarray, torch.Tensor],
patch_size: Union[int, Sequence[int]],
patch_overlap: Union[int, Sequence[int]],
):
self.array = data
patch_size = to_tuple(patch_size)
patch_overlap = to_tuple(patch_overlap)
self.locations = self._grid_spatial_coordinates(
self.array,
patch_size,
patch_overlap,
)
def parse_axes(axes: Union[int, Tuple[int, ...]]):
axes_tuple = to_tuple(axes)
for axis in axes_tuple:
is_int = isinstance(axis, int)
if not is_int or axis not in (0, 1, 2):
raise ValueError('All axes must be 0, 1 or 2')
return axes_tuple
def __init__(
self,
data: Union[torch.Tensor, np.ndarray],
patch_overlap: Union[int, Sequence[int]],
):
self.output_array = np.full(
data.shape,
fill_value=0,
dtype=np.uint16,
)
self.patch_overlap = to_tuple(patch_overlap)
def test_to_tuple(self):
assert to_tuple(1) == (1,)
assert to_tuple((1,)) == (1,)
assert to_tuple(1, length=3) == (1, 1, 1)
assert to_tuple((1, 2)) == (1, 2)
assert to_tuple((1, 2), length=3) == (1, 2)
assert to_tuple([1, 2], length=3) == (1, 2)
def test_to_tuple(self):
assert to_tuple(1) == (1, )
assert to_tuple((1, )) == (1, )
assert to_tuple(1, n=3) == (1, 1, 1)
assert to_tuple((1, 2)) == (1, 2)
assert to_tuple((1, 2), n=3) == (1, 2)
assert to_tuple([1, 2], n=3) == (1, 2)