point3d_rigid_match

Compute the correspondence between points in the two point sets before and after rigid transformation.

[!IMPORTANT] The algorithm in this project is only applicable to scenarios with nearly no noise, and it requires that the two point sets can be aligned via a single rotation plus a single translation. This algorithm is used to establish point correspondences when such correspondences are uncertain. It is only suitable for cases with a very small number of points (e.g., matching real-time coordinates of infrared optical tracking marker spheres with coordinates defined in the instrument description file).

Installation

pip install point3d_rigid_match

Usage

from point3d_rigid_match import find_correspondence

# Give two matrix of N * 3
# assume that they are generated by rigid transformation
A = [
    [-0.8583011244293051, 1.664188145937392, -5.255867327332927],
    [-0.10348646934862409, 2.087722438630186, -3.5977804293199576],
    [-0.037025558771713755, 1.502559878964906, -5.803026123816429],
    [1.314519244586987, 1.8135837080551638, -4.756535447571685]
]

B = [
    [0.9444699906120909, -0.00025015438771101525, 0.0018484750111346566],
    [-0.22389601437774168, -1.201344263998704, -0.0010553875203044244],
    [0.5505852240909018, 0.9189030480397202, -0.0016416782728876612],
    [-1.2711589253295885, 0.2826915078445261, 0.0008482241211740123]
]

# Output the correspondence between A and B
N = len(A)
corr = find_correspondence(A, B, rank_idx=0)
for i, j in enumerate(corr):
    print(f"A[{i}] -> B[{j}]")