estimate_covariances — 3D gicp op

Data kinds: pointsdescriptor

Call: import fullseye as fs; fs.ledger.estimate_covariances(points, k: 'int' = 20, epsilon: 'float' = 0.001) -> 'np.ndarray' (to call the implementation directly, import gicp; gicp.estimate_covariances(points, k: 'int' = 20, epsilon: 'float' = 0.001) -> 'np.ndarray'; from the registry, ops3d.get("estimate_covariances"))

Usage

Plane-to-plane covariances (N,3,3): each point's local covariance with its eigenvalues replaced by (ε,1,1).

> The detailed description below is the original text — the summary and the headings are translated.

k 近傍(自身を含む)の標本共分散を固有分解し、固有値を **接平面内は 1・

表面法線方向(最小固有値方向)は ε** に置き換えて再構成する。共分散の

大きさは近傍の広がりに依らず一定(ε,1,1)で、スケール不変な「面的な

確からしさ」モデルになる(GICP の中核)。

引数:

points (N,3): 点群。

k: 近傍数(自身を含む)。共分散推定は k≥4 程度が安定。

epsilon: 法線方向に割り当てる小分散(0<ε<1)。小さいほど接平面へ

強く拘束する。スケール不変(次元なし比)。

返り値:

(N,3,3) 対称正定値共分散の束。各共分散の固有値は {ε,1,1}、ε に

対応する固有ベクトルが局所表面法線方向。

例外:

ValueError: points が (N,3) でない / N<3(法線が定まらず縮退)/

epsilon が (0,1) 外(fail-closed)。

References (sample data, literature)

• 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.

Runnable examples (verified samples that actually call this op)

gicp_registerpy -3.11 examples_3d/gicp_register.py

Ops the type connects to (they accept descriptor as input)

fuse_to_voxel · shape_distance

Same category (gicp)

gicp


*Provenance: gicp.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.