simpleICP

This package contains an implementation of a rather simple version of the Iterative Closest Point (ICP) algorithm.

Documentation

This python implementation is just one of several (almost identical) implementations of the ICP algorithm in various programming languages. They all share a common documentation here: https://github.com/pglira/simpleICP

Installation

You can install the simpleicp package from PyPI:

pip install simpleicp

How to use

from simpleicp import PointCloud, SimpleICP
import numpy as np

# Read point clouds from xyz files into n-by-3 numpy arrays
X_fix = np.genfromtxt("bunny_part1.xyz")
X_mov = np.genfromtxt("bunny_part2.xyz")

# Create point cloud objects
pc_fix = PointCloud(X_fix, columns=["x", "y", "z"])
pc_mov = PointCloud(X_mov, columns=["x", "y", "z"])

# Create simpleICP object, add point clouds, and run algorithm!
icp = SimpleICP()
icp.add_point_clouds(pc_fix, pc_mov)
H, X_mov_transformed, rigid_body_transformation_params, distance_residuals = icp.run(max_overlap_distance=1)

This should give this output:

Consider partial overlap of point clouds ...
Select points for correspondences in fixed point cloud ...
Estimate normals of selected points ...
Start iterations ...
iteration | correspondences | mean(residuals) |  std(residuals)
   orig:0 |             863 |          0.0403 |          0.1825
        1 |             862 |          0.0096 |          0.1113
        2 |             775 |          0.0050 |          0.0553
        3 |             807 |          0.0022 |          0.0407
        4 |             825 |          0.0016 |          0.0346
        5 |             825 |          0.0010 |          0.0253
        6 |             816 |          0.0008 |          0.0198
        7 |             785 |         -0.0000 |          0.0142
        8 |             764 |          0.0008 |          0.0091
        9 |             753 |          0.0003 |          0.0061
       10 |             735 |          0.0002 |          0.0040
       11 |             742 |         -0.0001 |          0.0032
       12 |             747 |         -0.0000 |          0.0030
       13 |             752 |         -0.0000 |          0.0030
       14 |             752 |         -0.0000 |          0.0029
Convergence criteria fulfilled -> stop iteration!
Estimated transformation matrix H:
[    0.984798    -0.173702    -0.000053     0.000676]
[    0.173702     0.984798     0.000084    -0.001150]
[    0.000038    -0.000092     1.000000     0.000113]
[    0.000000     0.000000     0.000000     1.000000]
... which corresponds to the following rigid-body transformation parameters:
parameter |       est.value | est.uncertainty |       obs.value |      obs.weight
   alpha1 |       -0.004804 |        0.004491 |        0.000000 |       0.000e+00
   alpha2 |       -0.003061 |        0.002104 |        0.000000 |       0.000e+00
   alpha3 |       10.003124 |        0.005680 |        0.000000 |       0.000e+00
       tx |        0.000676 |        0.000418 |        0.000000 |       0.000e+00
       ty |       -0.001150 |        0.000885 |        0.000000 |       0.000e+00
       tz |        0.000113 |        0.000189 |        0.000000 |       0.000e+00
(Unit of est.value, est.uncertainty, and obs.value for alpha1/2/3 is degree)
Finished in 4.737 seconds!

Note that bunny_part1.xyz and bunny_part2.xyz are not included in this package. They can be downloaded (among other example files) here.