GeometryOperations#

class GeometryOperations[source]#

Bases: object

Registry for geometry operation functions.

Methods

`GeometryOperations.cluster`(geometry, method)	Partition points into clusters.
`GeometryOperations.compute_normals`(geometry)	Compute, flip, or remove point normals.
`GeometryOperations.downsample`(geometry[, method])	Downsample point cloud.
`GeometryOperations.duplicate`(geometry, **kwargs)	Duplicate a geometry.
`GeometryOperations.fit`(geometry, method, ...)	Fit a model to a point cloud.
`GeometryOperations.register`(operation_name, func)	Register an operation function.
`GeometryOperations.remesh`(geometry, method, ...)	Remesh by edge length, vertex count, or subdivision.
`GeometryOperations.remove_outliers`(geometry)	Remove statistical outliers from point cloud.
`GeometryOperations.sample`(geometry, ...[, ...])	Sample points from a fitted model.
`GeometryOperations.skeletonize`(geometry[, ...])	Extract structural skeleton from point cloud.
`GeometryOperations.smooth`(geometry, method, ...)	Smooth a triangle mesh.
`GeometryOperations.visibility`(geometry[, ...])	Change geometry visibility.

static cluster(geometry, method, drop_noise=False, use_points=True, use_normals=False, **kwargs)#

Partition points into clusters.

Parameters:

geometryGeometry or GeometryData: Input data.
methodstr: Clustering method to use. Options are: - ‘Connected Components’ : Connected components by spatial proximity - ‘Envelope’ : Envelope-based connected components - ‘Leiden’ : Leiden graph-based clustering - ‘DBSCAN’ : DBSCAN density-based clustering - ‘K-Means’ : K-Means partitioning - ‘Birch’ : Birch hierarchical clustering
drop_noisebool, optional: If True, drop noise points (label -1) from results. Default is False.
use_pointsbool, optional: If True, use point coordinates for clustering. Default is True.
use_normalsbool, optional: If True, include normal vectors in clustering features. Default is False.
**kwargs: Additional arguments passed to the chosen clustering method.

Returns:

List[mosaic.geometry.GeometryData]: List of geometries, one per cluster.

Raises:

ValueError: If unsupported clustering method is specified or too many clusters found.

static compute_normals(geometry, method='Compute', k=15, **kwargs)#

Compute, flip, or remove point normals.

Parameters:

geometryGeometry or GeometryData: Input data.
methodstr, optional: Normal computation method. Options are: - ‘Compute’ : Calculate new normals from point neighborhoods - ‘Flip’ : Reverse existing normal directions - ‘Remove’ : Clear any existing normal vectors Default is ‘Compute’.
kint, optional: Number of neighbors to consider for normal computation. Only used when method=’Compute’. Default is 15.
**kwargs: Additional parameters for normal computation.

Returns:

Geometry: A duplicate of the input geometry with updated normals.

static downsample(geometry, method='radius', **kwargs)#

Downsample point cloud.

Parameters:

geometryGeometry or GeometryData: Input data.
methodstr, optional: Method to use. Options are: - ‘radius’ : Uniform voxel grid downsampling - ‘center_of_mass’ : Replace nearby points by their centroid - ‘number’ : Random subsampling to target count Default is ‘radius’.
**kwargs: Additional arguments passed to the chosen method: - For ‘radius’: voxel_size parameter for open3d.voxel_down_sample - For ‘number’: size parameter specifying target number of points

Returns:

mosaic.geometry.GeometryData: Downsampled geometry.

static duplicate(geometry, **kwargs)#

Duplicate a geometry.

Parameters:

geometryGeometry or GeometryData: Geometry to duplicate.

Returns:

mosaic.geometry.GeometryData: Duplicated geometry.

static fit(geometry, method, **kwargs)#

Fit a model to a point cloud.

Parameters:

geometryGeometry or GeometryData: Input point cloud geometry.
methodstr: Fitting method. Options are: - ‘alpha_shape’ : Alpha-shape based surface reconstruction - ‘ball_pivoting’ : Ball pivoting surface reconstruction - ‘poisson’ : Poisson surface reconstruction - ‘flying_edges’ : Flying edges isosurface extraction - ‘shrink_wrap’ : Topology-preserving deformation (Reach for the Spheres) - ‘sphere’ : Least-squares sphere - ‘ellipsoid’ : Ellipsoid via eigenvalue decomposition - ‘cylinder’ : Cylinder via PCA and constrained optimization - ‘rbf’ : Radial basis function interpolation - ‘spline’ : Spline curve through control points

Returns:

mosaic.geometry.GeometryData: Geometry with fitted model attached.

Raises:

ValueError: If method is unsupported or point count is insufficient (<50).

classmethod register(operation_name, func, decorator=None)[source]#: Register an operation function.

static remesh(geometry, method, **kwargs)#

Remesh by edge length, vertex count, or subdivision.

Parameters:

geometryGeometry or GeometryData: Input geometry with a TriangularMesh model.
methodstr: Remeshing strategy. Options are: - ‘Decimation’ : Reduce triangle count via decimation - ‘Edge Length’ : Remesh to target edge length - ‘Subdivide’ : Subdivide mesh triangles - ‘Vertex Clustering’ : Simplify via vertex clustering

Returns:

mosaic.geometry.GeometryData or None: Remeshed geometry, or None if input has no mesh model.

static remove_outliers(geometry, method='statistical', **kwargs)#

Remove statistical outliers from point cloud.

Parameters:

geometryGeometry or GeometryData: Input data.
methodstr, optional: Outlier detection method. Options are: - ‘statistical’ : General statistical outlier removal - ‘eigenvalue’ : Eigenvalue-based edge noise removal Default is ‘statistical’.
**kwargs: Additional parameters for outlier removal method.

Returns:

mosaic.geometry.GeometryData or None: Filtered point cloud geometry with outliers removed. Returns None if no points remain after filtering.

static sample(geometry, sampling, method, normal_offset=0.0, bidirectional=False, **kwargs)#

Sample points from a fitted model.

Parameters:

geometryGeometry or GeometryData: Input data.
samplingfloat: Sampling rate or number of points to generate.
methodstr: Sampling method to use: - ‘distance’ : Sample by average inter-point distance - ‘points’ : Sample by total point count
normal_offsetfloat, optional: Point offset along normal vector, defaults to 0.0.
bidirectionalbool, optional: Draw inward and outward facing points at the same time. This doubles the total number of points. Default is False.

Returns:

mosaic.geometry.GeometryData: Sampled geometry.

Raises:

ValueError: If geometry has no fitted model.

static skeletonize(geometry, method='core', sigma=1.0, **kwargs)#

Extract structural skeleton from point cloud.

Parameters:

geometryGeometry or GeometryData: Input data.
method{‘outer’, ‘core’, ‘boundary’}, optional: Structural feature to extract: - ‘core’ : Extract medial axis/centerline through the middle of structures - ‘boundary’ : Extract both inner and outer boundaries for hollow structures - ‘outer’ : Extract outer boundary via skeletonization + convex hull - ‘outer_hull’ : Fast convex hull approximation (no skeletonization)
sigmafloat, optional: Gaussian smoothing for Hessian computation.
**kwargs: Additional arguments passed to the chosen method.

Returns:

mosaic.geometry.GeometryData: Skeletonized point cloud.

Raises:

ValueError: If unsupported method is specified.

static smooth(geometry, method, **kwargs)#

Smooth a triangle mesh.

Parameters:

geometryGeometry or GeometryData: Input geometry with a TriangularMesh model.
methodstr: Smoothing algorithm. Options are: - ‘Taubin’ : Volume-preserving Taubin smoothing - ‘Laplacian’ : Laplacian mesh smoothing - ‘Average’ : Simple neighbor averaging - ‘Fair’ : Polyharmonic mesh deformation

Returns:

mosaic.geometry.GeometryData or None: Smoothed geometry, or None if input has no mesh model.

static visibility(geometry, visible=True, **kwargs)#

Change geometry visibility.

Parameters:

geometryGeometry or GeometryData: Target geometry.
visiblebool, optional: Whether the geometry should be visible. Default is True.