fit op• Data kinds: keypoints → table
• Call: import fullseye as fs; fs.ledger.fit_curve_principal_axis(points, degree: 'int' = 2, *, isotropy_ratio: 'float' = 0.95) -> 'dict' (to call the implementation directly, import measure; measure.fit_curve_principal_axis(points, degree: 'int' = 2, *, isotropy_ratio: 'float' = 0.95) -> 'dict'; from the registry, opsfit.get("fit_curve_principal_axis"))
• GPU: none (CPU reference implementation only)
Rotate a slanted curved edge to its principal axis, then fit polynomial y'=f(x').
Returns the image-plane angle_deg, centroid (row, col), polynomial coefficients in NumPy's descending-power convention, orthonormal axes (rows are local x/y axes in col,row order), local coordinates, fitted_points in the original row/col frame, and a gradient-normalised geometric rms. degree must be at least one and leave more samples than coefficients. A nearly round cloud (minor/major variance ratio at or above isotropy_ratio) is rejected because its principal direction is undefined.
• 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.
• The canonical algorithm (author, year) and its uses are named in the family usage guide above.
• fit_ops_from_knowhow — py -3.11 examples/fit_ops_from_knowhow.py
table as input)—
fit)*Provenance: measure.py — FIT 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.