morphology op• Data kinds: voxel → voxel
• Call: import fullseye as fs; fs.ledger.morph_dilate3d(vol, r=1, device='cpu', se='cube') (to call the implementation directly, import match3d; match3d.morph_dilate3d(vol, r=1, device='cpu', se='cube'); from the registry, ops3d.get("morph_dilate3d"))
• GPU: this op has a GPU path (device="cuda")
3-D greyscale dilation (the local max over a structuring element of radius r). Expands the bright regions.
> The detailed description below is the original text — the summary and the headings are translated.
se="cube"(既定、torch 経路で GPU 可)/ "ball"(等方 SE、scipy 経路)。
SE は一辺 `2r+1 の立方体(cube)か z²+y²+x² <= r² の球(ball)。r=0` は恒等。
境界の外は −∞ 扱い(画像内の値だけで max を取る。torch の `max_pool3d` の implicit
padding と scipy の `cval=-inf で同じ結果)。cube は torch があれば max_pool3d`
(`device 有効)、ball または torch 不在なら scipy.ndimage.grey_dilation(device` は
無視)。入力は float32 に変換され、返り値 `(D,H,W) float32 numpy。se` がその 2 つ以外なら
ValueError。2 値 volume(0/1)ならそのまま 2 値 dilation になる。`se="ball"` は r が
大きいと footprint 走査で遅い。
後段: `morph_erode3d と組で morph_open3d / morph_close3d / morph_gradient3d`。
• 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.
• morphology_3d — py -3.11 examples_3d/morphology_3d.py
voxel as input)voxel_to_mips · voxel_to_mesh · signed_distance_field · to_points · sobel3d · hessian3d · curvature_maps · edt_jfa
morphology)morph_erode3d · morph_open3d · morph_close3d · morph_gradient3d · morph_tophat3d · morph_blackhat3d
*Provenance: match3d.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.