miop.matrices_metadata

Classes

CameraMvmtFromMetadata()

Computes camera intrinsics and extrinsics (rotation matrices) from metadata.
class miop.matrices_metadata.CameraMvmtFromMetadata[source]

Bases: DAGNode

Computes camera intrinsics and extrinsics (rotation matrices) from metadata.

rotation_matrices

Rotation matrices for each image in each face.

Type:

list of list of np.ndarray

camera_position
List of dictionaries for each face containing:

  • ‘rotations’: array of 3x3 rotation matrices (extrinsics)

  • ‘intrinsics’: placeholder list for intrinsics (scaling and shear [ones, [1., 0.]])

Type:

list of dict

choose_axis(rot_z)[source]

Determines the tilt axis based on rotation in the (x, y) plane.

Parameters:

rot_z (float) – Rotation angle in the (x, y) plane in radians.

Returns:

A 3D unit vector representing the axis of rotation, one of: - [1., 0., 0.] (default: horizontal) - [-1., 0., 0.] (rot_z ≈ 180°) - [0., 1., 0.] (rot_z ≈ -90°) - [0., -1., 0.] (rot_z ≈ 90°)

Return type:

list of float

eval(img_collection)[source]

Computes camera extrinsics (rotation matrices) for each image in the image collection using the tilt angle and inferred tilt axis from metadata.

Parameters:

img_collection (ImageCollection) – The collection of images with metadata from which to compute rotations.

Returns:

One dictionary per face, containing:

  • ’rotations’: np.ndarray of shape (N, 3, 3), rotation matrices for N images

  • ’intrinsics’: placeholder value ([ones, [1., 0.]])

Return type:

list of dict

get_rotation_matrix(angle, axis)[source]

Computes a 3x3 rotation matrix given an angle and rotation axis using Rodrigues’ formula.

Parameters:

  • angle (float) – Rotation angle in radians.

  • axis (list of float) – 3D axis of rotation (e.g., [1., 0., 0.] for X-axis).

Returns:

A 3x3 rotation matrix representing the rotation about the given axis.

Return type:

np.ndarray