OpNavManager

class OpNavManager

Bases: object

Pinhole camera projection shared by the optical measurement models.

A stateless collection of the calibration matrix, the pinhole projection and its Jacobian, so the projection and its derivative have exactly one implementation and stay in step.

All methods are class or static methods; the class is never instantiated.

Methods

calibration_matrix(camera)	Return the pinhole calibration matrix K of a camera.
project(camera, points_camera)	Pinhole-project (N, 3) camera-frame vectors to (N, 2) pixels.
project_jacobian(camera, point_camera)	Return d(u, v) / d r_cam — the pinhole projection Jacobian.

See also

scarabaeus.CentroidingIdeal

Centroid model built on this projection.

classmethod project(camera, points_camera: ndarray) → ndarray

Pinhole-project (N, 3) camera-frame vectors to (N, 2) pixels.

Parameters:

cameraCamera

The camera supplying the calibration matrix.

points_cameranumpy.ndarray

(N, 3) vectors from the camera to the points, in the camera frame.

Returns:

numpy.ndarray

(N, 2) pixel coordinates.

classmethod project_jacobian(camera, point_camera: ndarray) → ndarray

Return d(u, v) / d r_cam — the pinhole projection Jacobian.

Differentiates project() for a single point: \(u = f_x x/z + c_x\) and \(v = f_y y/z + c_y\), so

\[\begin{split}\frac{\partial(u, v)}{\partial\mathbf{r}_{\text{cam}}} = \begin{pmatrix} f_x/z & 0 & -f_x x/z^2 \\ 0 & f_y/z & -f_y y/z^2 \end{pmatrix}\end{split}\]

Parameters:

cameraCamera

The camera supplying the calibration matrix.

point_cameranumpy.ndarray

(3,) vector from the camera to the point, in the camera frame.

Returns:

numpy.ndarray

(2, 3) Jacobian, or zeros when the point lies in the focal plane (z == 0) and the projection is undefined.

static calibration_matrix(camera) → ndarray

Return the pinhole calibration matrix K of a camera.

Camera.calibration_matrix is stored as the (2, 3) projection block [[fx, 0, cx], [0, fy, cy]]; the homogeneous row is appended here.

Parameters:

cameraCamera

The camera whose intrinsics are wanted.

Returns:

numpy.ndarray

(3, 3) calibration matrix.