ImageDataFilters.label_connectivity

ImageDataFilters.label_connectivity(

*,

scalars: str | None = None,

scalar_range: Literal['auto', 'foreground', 'vtk_default'] | VectorLike[float] = 'auto',

extraction_mode: Literal['all', 'largest', 'seeded'] = 'all',

point_seeds: MatrixLike[float] | VectorLike[float] | _vtk.vtkDataSet | None = None,

label_mode: Literal['size', 'constant', 'seeds'] = 'size',

constant_value: int | None = None,

inplace: bool = False,

progress_bar: bool = False,

) → tuple[ImageData, NDArray[int], NDArray[int]]

Find and label connected regions in a ImageData.

Only points whose scalar value is within the scalar_range are considered for connectivity. A 4-connectivity is used for 2D images or a 6-connectivity for 3D images. This filter operates on point-based data. If cell-based data are provided, they are re-meshed to a point-based representation using cells_to_points() and the output is meshed back to a cell-based representation with points_to_cells(), effectively filtering based on face connectivity. The connected regions are extracted and labelled according to the strategy defined by extraction_mode and label_mode, respectively. Unconnected regions are labelled with 0 value.

Added in version 0.45.0.

Parameters:

scalarsstr, optional

Scalars to use to filter points. If None is provided, the scalars is automatically set, if possible.

scalar_rangestr, Literal[‘auto’, ‘foreground’, ‘vtk_default’], VectorLike[float]

Points whose scalars value is within 'scalar_range' are considered for connectivity. The bounds are inclusive.

• 'auto': (default) includes the full data range, similarly to connectivity().

• 'foreground': includes the full data range except the smallest value.

• 'vtk_default': default to [0.5, VTK_DOUBLE_MAX].

• VectorLike[float]: explicitly set the range.

The bounds are always cast to floats since vtk expects doubles. The scalars data are also cast to floats to avoid unexpected behavior arising from implicit type conversion. The only exceptions is if both bounds are whole numbers, in which case the implicit conversion is safe. It will optimize resources consumption if the data are integers.

extraction_modeLiteral[‘all’, ‘largest’, ‘seeded’], default: ‘all’

Determine how the connected regions are extracted. If 'all', all connected regions are extracted. If 'largest', only the largest region is extracted. If 'seeded', only the regions that include the points defined with point_seeds are extracted.

point_seedsMatrixLike[float], VectorLike[float], vtkDataSet, optional

The point coordinates to use as seeds, specified as a (N, 3) array like or as a vtkDataSet. Has no effect if extraction_mode is not 'seeded'.

label_modeLiteral[‘size’, ‘constant’, ‘seeds’], default: ‘size’

Determine how the extracted regions are labelled. If 'size', label regions by decreasing size (i.e., count of cells), starting at 1. If 'constant', label with the provided constant_value. If 'seeds', label according to the seed order, starting at 1.

constant_valueint, optional

The constant label value to use. Has no effect if label_mode is not 'seeds'.

inplacebool, default: False

If True, perform an inplace labelling of the ImageData. Else, returns a new ImageData.

progress_barbool, default: False

Display a progress bar to indicate progress.

Returns:

pyvista.ImageData

Either the input ImageData or a generated one where connected regions are labelled with a 'RegionId' point-based or cell-based data.

NDArray[int]

The labels of each extracted regions.

NDArray[int]

The size (i.e., number of cells) of each extracted regions.

See also

pyvista.DataSetFilters.connectivity

Similar general-purpose filter that performs 1-connectivity.

Notes

This filter implements vtkImageConnectivityFilter.

Examples

Prepare a segmented grid.

>>> import pyvista as pv
>>> segmented_grid = pv.ImageData(dimensions=(4, 3, 3))
>>> segmented_grid.cell_data['Data'] = [
...     0,
...     0,
...     0,
...     1,
...     0,
...     1,
...     1,
...     2,
...     0,
...     0,
...     0,
...     0,
... ]
>>> segmented_grid.plot(show_edges=True)

Static Scene

Interactive Scene

Label the connected regions. The cells with a 0 value are excluded from the connected regions and labelled with 0. The remaining cells define 3 different regions that are labelled by decreasing size.

>>> connected, labels, sizes = segmented_grid.label_connectivity(
...     scalar_range='foreground'
... )
>>> pl = pv.Plotter()
>>> _ = pl.add_mesh(connected.threshold(0.5), show_edges=True)
>>> _ = pl.add_mesh(
...     connected.threshold(0.5, invert=True),
...     show_edges=True,
...     opacity=0.5,
... )
>>> pl.show()

Static Scene

Interactive Scene

Exclude the cell with a 2 value.

>>> connected, labels, sizes = segmented_grid.label_connectivity(
...     scalar_range=[1, 1]
... )
>>> pl = pv.Plotter()
>>> _ = pl.add_mesh(connected.threshold(0.5), show_edges=True)
>>> _ = pl.add_mesh(
...     connected.threshold(0.5, invert=True),
...     show_edges=True,
...     opacity=0.5,
... )
>>> pl.show()

Static Scene

Interactive Scene

Label all connected regions with a constant value.

>>> connected, labels, sizes = segmented_grid.label_connectivity(
...     scalar_range='foreground',
...     label_mode='constant',
...     constant_value=10,
... )
>>> pl = pv.Plotter()
>>> _ = pl.add_mesh(connected.threshold(0.5), show_edges=True)
>>> _ = pl.add_mesh(
...     connected.threshold(0.5, invert=True),
...     show_edges=True,
...     opacity=0.5,
... )
>>> pl.show()

Static Scene

Interactive Scene

Label only the regions that include seed points, by seed order.

>>> points = [(2.0, 1.0, 0.0), (0.0, 0.0, 1.0)]
>>> connected, labels, sizes = segmented_grid.label_connectivity(
...     scalar_range='foreground',
...     extraction_mode='seeded',
...     point_seeds=points,
... )
>>> pl = pv.Plotter()
>>> _ = pl.add_mesh(connected.threshold(0.5), show_edges=True)
>>> _ = pl.add_mesh(
...     connected.threshold(0.5, invert=True),
...     show_edges=True,
...     opacity=0.5,
... )
>>> pl.show()

Static Scene

Interactive Scene