calibrate op• 数据种类:keypoints → table
• 调用: import fullseye as fs; fs.ledger.stereo_calibration_lm(object_points, image_points_left, image_points_right, dist_model='k1k2p1p2k3', fix_intrinsics=False, calib_left=None, calib_right=None, image_shape=None)(要直接调用实现,import calib_lm; calib_lm.stereo_calibration_lm(object_points, image_points_left, image_points_right, dist_model='k1k2p1p2k3', fix_intrinsics=False, calib_left=None, calib_right=None, image_shape=None);从台账取用则 opscalib.get("stereo_calibration_lm"))
• GPU:无(仅 CPU 参考实现)
用 LM 同时优化左右相机的内参与畸变、板的位姿以及右相机的相对位姿 (R, T)(table)。
> 以下的详细说明为原文 —— 摘要与标题已翻译。
Joint stereo calibration: both cameras' intrinsics and distortion, the board poses in the left camera and the right camera's pose relative to the left (X_right = R·X_left + T) are refined together by Levenberg-Marquardt on the reprojection error in both images. Each camera starts from camera_calibration_lm (or calib_left / calib_right if given) and (R, T) from the per-view relative poses (rotation by the chordal mean, translation by the mean). fix_intrinsics keeps both cameras' intrinsics and distortion fixed (OpenCV's CALIB_FIX_INTRINSIC).
返す dict: K1・dist1・K2・dist2・R・T・E = [T]x R・F = K2⁻ᵀ E K1⁻¹ (画素 (u = col, v = row) の同次座標どうし)、baseline = |T|、reproj_rms(左右合わせて)、rms_left・rms_right、poses_left、sigma_px、rel_std(R の回転ベクトルと T の標準偏差 6 個)、iterations・converged。
Raises ValueError: 左右の視点の数が違う・3 未満、点の数が合わない、単眼の校正の失敗。
• 示例数据目录(下载 URL / 许可证) —— 2-D 用 skimage.data(BSD/公有领域)加合成图,3-D 给出真实数据源(Stanford/PDS 等)的下载 URL。
• 算子来历与参考文献 —— 该算子族所依据的研究/方法出处。
• 算法的正典(作者、年份)与用途见上面的族使用指南。
• (暂无)
table 作为输入)image_points_to_world_plane · contour_to_world_plane_xld · sim_caltab · disp_caltab · calibration_error_map · calibration_uncertainty_map · calibration_next_pose
calibrate)camera_calibration · find_marks_and_pose · hand_eye_calibration · camera_calibration_lm
*Provenance: calib_lm.py — CALIB 算子登记表。本条目由 tools/opdocs.py md 自动生成(请勿手工编辑)。*
© 2026 Kazufumi Furuse — Fullseye operator documentation. Licensed under Apache-2.0.