Connectivity

Connectivity#

Download Python source code | Download Jupyter notebook

Highlight some applications of the connectivity() filter.

Remove Noisy Isosurfaces#

Use connectivity to remove noisy isosurfaces.

This section is similar to this VTK example.

import numpy as np
import pyvista as pv
from pyvista import examples

Load a dataset with noisy isosurfaces.

pine_roots = examples.download_pine_roots()

# Plot the raw data
cpos = pv.CameraPosition(
    position=(40.6, -280.5, 47.02),
    focal_point=(40.6, 37.28, 50.2),
    viewup=(0.0, 0.0, 1.0),
)
pl = pv.Plotter()
pl.add_mesh(pine_roots, color='#965434')
pl.add_mesh(pine_roots.outline())
pl.show(cpos=cpos)

Static Scene

Interactive Scene

The plotted mesh is very noisy. We can extract the largest connected isosurface using the 'largest' extraction_mode of the connectivity() filter. Equivalently, extract_largest() may also be used.

# Grab the largest connected volume present
largest = pine_roots.connectivity('largest')
# or: largest = mesh.extract_largest()

pl = pv.Plotter()
pl.add_mesh(largest, color='#965434')
pl.add_mesh(pine_roots.outline())
pl.camera_position = cpos
pl.show()

Static Scene

Interactive Scene

Extract Small Regions#

Use connectivity to extract the smaller ‘noisy’ regions that were removed in the remove noisy isosurfaces example above.

First, get a list of all region ids.

all_regions = pine_roots.connectivity('all')
region_ids = np.unique(all_regions['RegionId'])

Since the region IDs are sorted in descending order (by cell count), we can extract all regions except for the largest one using the 'specified' extraction_mode of the connectivity() filter.

noise_region_ids = region_ids[1::]  # All region ids except '0'
noise = pine_roots.connectivity('specified', noise_region_ids)

Plot the noisy regions using color_labels(). For context, also plot the largest region.

pl = pv.Plotter()
pl.add_mesh(noise.color_labels())
pl.add_mesh(largest, color='lightgray', opacity=0.1)
pl.add_mesh(pine_roots.outline())
pl.camera_position = cpos
pl.show()

Static Scene

Interactive Scene

Label Disconnected Regions#

Use connectivity to label all disconnected regions.

This section is similar to this VTK example.

First, load a dataset with disconnected regions.

mesh = examples.download_foot_bones()

Extract all regions.

conn = mesh.connectivity('all')

Plot the labeled regions using color_labels().

colored = conn.color_labels()
cpos = pv.CameraPosition(
    position=(10.5, 12.2, 18.3), focal_point=(0.0, 0.0, 0.0), viewup=(0.0, 1.0, 0.0)
)
colored.plot(cpos=cpos)

Static Scene

Interactive Scene

Extract Regions From Seed Points#

Use connectivity to extract regions of interest using scalar data and seed points.

First, create a dataset with salient features. Here, we create hills and use curvature to define its peaks and valleys.

mesh = pv.ParametricRandomHills()
mesh['Curvature'] = mesh.curvature()

Visualize the peaks and valleys. Peaks have large positive curvature (that is, they are convex). Valleys have large negative curvature (that is, they are concave). Flat regions have curvature close to zero.

mesh.plot(
    clim=[-0.5, 0.5],
    cmap='bwr',
    below_color='blue',
    above_color='red',
)

Static Scene

Interactive Scene

Extract a region of interest using the 'point_seed' extraction_mode of the connectivity() filter. Let’s extract the steepest peak using a seed point where the curvature is maximized.

# Get seed point
peak_point_id = np.argmax(mesh['Curvature'])

# Define a scalar range of the region to extract
data_min, data_max = mesh.get_data_range()
peak_range = [0.2, data_max]  # Peak if curvature > 0.2

peak_mesh = mesh.connectivity('point_seed', peak_point_id, scalar_range=peak_range)

Let’s also extract the closest valley to the steepest peak using the 'closest' extraction_mode of the connectivity() filter.

valley_range = [data_min, -0.2]  # Valley if curvature < -0.2
peak_point = mesh.points[peak_point_id]
valley_mesh = mesh.connectivity('closest', peak_point, scalar_range=valley_range)

Plot extracted regions.

pl = pv.Plotter()
_ = pl.add_mesh(mesh, style='wireframe', color='lightgray')
_ = pl.add_mesh(peak_mesh, color='red', label='Steepest Peak')
_ = pl.add_mesh(valley_mesh, color='blue', label='Closest Valley')
_ = pl.add_legend()
pl.show()