Estimates a camera's intrinsics (focal lengths, principal point) and lens distortion from several chessboard views (cv2.calibrateCamera). Calibration needs per-view 2D/3D correspondences accumulated across images, which the raw wrappers can't express, so this node manages it: feed an IMAGE BATCH of board photos from DIFFERENT angles; each is grayscaled, its inner corners are found and refined to sub-pixel (cornerSubPix), and the matching planar object points are built from pattern size x square_size. Views where the board is not found are skipped. Failure-tolerant: with fewer than 3 usable views it returns found=false, an identity matrix and zero distortion. Feed camera_matrix + dist_coeffs into cv2.undistort, or cv2.solvePnP for pose; rms_error (px) reports fit quality (under ~1 is good).
By bmad4ever·Created 3 months ago·Updated 2 days ago· 0
CV Calibrate Camera (Chessboard)
images
camera_matrix
dist_coeffs
rms_error
views_used
found
◄pattern_cols9►
◄pattern_rows6►
◄square_size1.00►
Categoryimage/CV/features
Inputs (4)
Name
Type
Default
Description
images
IMAGE
Batch of chessboard views from different angles (>= 3 usable; ~10-20 gives a good fit).
pattern_cols
INT
92–40
Inner corners per row (squares per row minus 1).
pattern_rows
INT
62–40
Inner corners per column (squares per column minus 1).
square_sizeopt
FLOAT
1.000.0001–1000000
Physical size of one square (e.g. mm); sets the world scale. Leave 1.0 for relative calibration.
Outputs (5)
Name
Type
Description
camera_matrix
NPARRAY
3x3 intrinsic matrix K (fx, fy, cx, cy).
dist_coeffs
NPARRAY
Distortion coefficients (k1, k2, p1, p2, k3).
rms_error
FLOAT
Mean reprojection error in pixels; lower is better (<1 good).
