calib_truth op• Data kinds: table → table
• Call: import fullseye as fs; fs.ledger.calibration_uncertainty_map(calibration, image_shape, step=16, depth=None, remove_rotation=True) (to call the implementation directly, import calib_lm; calib_lm.calibration_uncertainty_map(calibration, image_shape, step=16, depth=None, remove_rotation=True); from the registry, opscalib.get("calibration_uncertainty_map"))
• GPU: none (CPU reference implementation only)
Maps the covariance of the calibration estimate onto the image and returns the predicted uncertainty of every pixel (px, or mm at depth Z) (table).
> The detailed description below is the original text — the summary and the headings are translated.
Propagate the marginal covariance of the intrinsics and distortion (from camera_calibration_lm) to every grid pixel: a point fixed in the camera frame that the estimate places at p would be reprojected with a standard deviation sqrt(trace(J C Jᵀ)) (px, radial), J the 2x9 Jacobian of the projection with respect to (fx, fy, cx, cy, k1, k2, p1, p2, k3). Needs no ground truth — this is the map a real calibration can show too; the simulation checks that it is honest (Monte Carlo gate). With depth Z (mm) the map is also given in mm (std · Z / f, small-angle).
remove_rotation(既定 True)は calibration_error_map と同じ意味: カメラの向きのずれと見分けのつかない成分(主点のずれの大部分)を除く。両方を同じ設定で使うと、モンテカルロの誤差の地図と比べられる。★姿勢の不確かさは周辺化で含まれる(9x9 は全変数の共分散からの周辺)。画像の端ほど大きくなるのは、板が写っていない所を歪みの多項式が外挿するから。★線形の予測なので、外挿が大きい所では当たらない:
6 視点・σ = 0.2 px で k3 まで自由にすると、隅の予測 160 px に対しモンテカルロの実測は 55 px (k1k2p1p2 なら全体の平均で 6.42 / 6.39 px と合う。tests の実測)。
返す dict: rows・cols、std_px(地図)、mean_px・max_px・max_at、std_mm(depth)。
Raises ValueError: 共分散が無い・非有限、引数の範囲外。
• camera_calibration family guide
• 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.
• poc_calibration_truth_map — py -3.11 examples/poc_calibration_truth_map.py
table as input)image_points_to_world_plane · contour_to_world_plane_xld · sim_caltab · disp_caltab · calibration_error_map · calibration_next_pose
calib_truth)calibration_error_map · calibration_next_pose
*Provenance: calib_lm.py — CALIB 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.