mymesh.register.Mesh2Mesh

mymesh.register.Mesh2Mesh(M1, M2, T0=None, bounds=None, transform='rigid', metric='icp', method='icp', decimation=1, transform_args={}, optimizer_args=None, verbose=True)

Mesh-to-mesh registration. M2 will be aligned to M1.

Parameters:

• M1 (mymesh.mesh) – Fixed mesh that M2 will be aligned to

• M2 (mymesh.mesh) – Moving mesh that will be aligned to M1

• T0 (array_like or NoneType, optional) – Initial transformation to apply to M2, by default, None. This can serve as an ‘initial guess’ of the alignment.

• bounds (array_like or NoneType, optional) – Optimization bounds, formatted as [(min,max),…] for each parameter. If None, bounds are selected that should cover most possible transformations, by default None. Not used by all optimizers.

• transform (str, optional) –

  Transformation model, by default ‘rigid’.

  • ’rigid’: translations and rotations

  • ’similarity: translations, rotations, and uniform scaling

  • ’affine’: translations, rotations, triaxial scaling, shearing

• metric (str, optional) – Similarity metric to compare the two point clouds, by default ‘closest_point_MSE’

• method (str, optional) – Optimization method, by default ‘direct’. See optimize() for details.

• decimation (float, optional) – Scalar factor in the range (0,1] used to reduce the size of the point set, by default 1. decimation = 1 uses the full point sets, numbers less than one will reduce the size of both point sets by that factor by randomly selecting a set of points to use. For example decimation = 0.5 will use only half of the points of each set. A random seed of 0 is used for repeatable results. Note that if verbose=True, the final score will be reported for the full point set, not the decimated point set used during optimization.

• transform_args (dict, optional) – Additional arguments for the chosen transformation model, by default {}. See rigid(), similarity(), or affine().

• optimizer_args (dict, optional) – Additional arguments for the chosen optimizer, by default None. See optimize() for details.

• verbose (bool, optional) – Verbosity, by default True. If True, iteration progress will be printed.

Returns:

• Mnew (mymesh.mesh) – Transformed transformed mesh registered to M1.

• T (np.ndarray) – Affine transformation matrix (shape=(4,4)) to transform M2 to Mnew. Mnew.NodeCoords = transform_points(M.NodeCoords, T)

Examples

import mymesh
from mymesh import register
import numpy as np

# Load two versions of the stanford bunny (fine and coarsened)
m1 = mymesh.demo_mesh('bunny')
m2 = mymesh.demo_mesh('bunny-res2')

# Perform an arbitrary rotation to the copy
R = register.rotation([np.pi/6, -np.pi/6, np.pi/6])
m2.NodeCoords = register.transform_points(m2.NodeCoords, R)

# Align the two meshes using the iterative closest point (ICP) algorithm
m_aligned, T = register.Mesh2Mesh(m1, m2, method='icp')