Calibration(files, n_corners, checker_size, extrinsic_index)

Build a calibration object. files are the image files of the checkerboard. n_corners is a tuple of the number of corners in each of the sides of the checkerboard. checker_size is the physical size of the checker (e.g. in cm). with_distortion controls if radial lens distortion is included in the model or not.

c(i::RowCol, extrinsic)

Convert the row column StaticArray i to its real-world equivalent, XYZ, for the extrinsic parameters from the extrinsic image (given as an index or file name).

c(xyz::XYZ, extrinsic)

Convert the x, y, z StaticArray xyz to its pixel-coordinate equivalent, RowCol for the extrinsic parameters from the extrinsic image (given as an index or file name).

RowCol(row, col)

An alias for a static vector of two, row and column, indicating a cartesian coordinate in an image/matrix.

XYZ(x, y, z)

An alias for a static vector of three, x, y, and z, indicating a real-world coordinate. Note that x is equivalent to the column in RowCol and the y is equivalent to the row.

_detect_corners

Wraps OpenCV function to auto-detect corners in an image.

calculate_errors(c)

Calculate reprojection, projection, distance, and inverse errors for the calibration c. distance measures the mean error of the distance between all adjacent checkerboard corners from the expected checker_size. inverse measures the mean error of applying the calibration's transformation and its inverse inverse_samples times.

fit_model

Wraps OpenCV function to fit a camera model to given object and image points.

get_object_points

Produce the real-world locations of the corners of the checkerboard.

improve

Identify all the images that had relatively high reprojection errors, and rerun the calibration without them. Include a maximum of n images with the lowest reprojection error, or all the images with an error lower than threshold.

inv_lens_distortion

Analytical inverse lens distortion for one radial coefficient.

lens_distortion

Lens distortion for one radial coefficient.

plot

Save to locally created directory debug all the calibration's images after rectification. Blue crosses indicate detetcted corners, and red crosses indicate their rectificated real-world locations. Ideally, both crosses' centers are right on top of each other.

rectification(c, extrinsic_index)

Return a function that accepts an instance of ::RowCol and converts it to its real-world equivalent for the extrinsic parameters from the extrinsic_index image, without its third dimension (which would be ≈ 0): an xy coordinate.