feature op• Data kinds: voxel → voxel
• Call: import fullseye as fs; fs.ledger.vol_local_thickness(vol_binary, spacing=None, max_radius=None) (to call the implementation directly, import volops; volops.vol_local_thickness(vol_binary, spacing=None, max_radius=None); from the registry, ops3d.get("vol_local_thickness"))
Local thickness map: the diameter of the largest ball that covers each voxel.
For every foreground voxel this is ``max{ 2r : the voxel lies inside some ball
of radius r that fits entirely in the foreground }`` — the classical
granulometric size, and the quantity trabecular-bone and industrial-CT work
calls *local thickness* (a pore's local thickness is its diameter; a wall's is
its thickness). Computed from the exact distance transform: take the voxels
whose distance is at least `r` as ball centres and paint the balls they
cover, largest radius first.
★**Distinct from `vol_wall_thickness`**, which walks a single probe segment
`p0 -> p1` and returns a list of crossings. This one is a *volume*, so the
thickness of every feature is available at once and can be histogrammed.
Pass *spacing* `(sz, sy, sx) (or a :class:volio.VolumeMeta`) and the
result is in millimetres; otherwise in voxels. *max_radius* caps the
search (in the same units); leave it `None` to derive it from the largest
distance actually present, which costs one pass more but never saturates.
Applicability. (1) The foreground comes from a single threshold, so
beam hardening or a brightness gradient biases the thickness across the
volume — flatten first. (2) With anisotropic voxels the ball is a ball in
millimetres, not in voxels, so *spacing* is not cosmetic. (3) A feature
three voxels across is quantised at better than 30% only if you say so:
report the voxel size next to any thickness number.
Returns a `(D, H, W)` float64 volume, 0 on the background.
• Sample-data catalog (download URLs / licences) — 2-D uses skimage.data (BSD/public domain) plus synthetic images; 3-D lists download URLs for real data sources (Stanford, PDS, …).
• Operator provenance and references — the sources of the research/methods this op family came from.
• ct_porosity_and_fibre_morphometry — py -3.11 examples_3d/ct_porosity_and_fibre_morphometry.py
voxel as input)voxel_to_mips · voxel_to_mesh · signed_distance_field · to_points · sobel3d · hessian3d · curvature_maps · edt_jfa
feature)sobel3d · hessian3d · curvature_maps · edt_jfa · vol_frangi · vol_local_std · vol_orientation_coherence · vol_euler_number
*Provenance: volops.py — 3D operator registry. This per-op note is generated by tools/opdocs.py md (do not hand-edit).*
© 2026 Kazufumi Furuse — Fullseye operator documentation. Licensed under Apache-2.0.