Normal estimation for point clouds

Usage

nrml_estimator = PointCloudNormalEstimator([options])
nrml_estimation.run(mesh)

or

nrml_estimator = PointCloudNormalEstimator([options])(mesh) # directly calls run

This fills the various container attributes that can be accessed in the PointCloudNormalEstimator class.

A point cloud of the spot model, which is a little cow, where each point is associated an outward normal vector — Outward normal estimation on a point cloud of the spot model

PointCloudNormalEstimator

Bases: Worker

An estimator of normal directions for unstructured point clouds, based on k-nearest neighbors and singular value decomposition of their correlation matrix.

References

[1] Surface reconstruction from unorganized points, Hoppe et al., 1992

Example

https://github.com/GCoiffier/mouette/blob/main/examples/point_cloud_normal_estimation.py

Parameters:

Name	Type	Description	Default
`n_neighbors`	`int`	number of neighbor points to consider for local plane fitting. Defaults to 5.	`5`
`save_on_pc`	`bool`	whether to store the resulting normals onto the point cloud as a vertex attribute, or as independent arrays. Defaults to True.	`True`
`compute_curvature`	`bool`	whether to also compute an estimate of curvature on the point cloud. This curvature estimate is computed as the ratio of the smallest eigenvalue and the sum of eigenvalues of the correlation matrix at each point (between 0 and 1). Defaults to False.	`False`
`orientation_mode`	`str`	Heuristic strategy for consistent orientation of normals. Choices are ["None", "mst"] Defaults to "mst".	`'mst'`
`verbose`	`bool`	verbose mode. Defaults to True.	`True`

Other Parameters:

Name	Type	Description
`normal_attribute_name`	`str`	The name of the attribute in which normals are stored. Ignored if save_on_pc is set to False. Defaults to "normals".
`curvature_attribute_name`	`str`	The name of the attribute in which the curvature estimation is stored. Ignored if save_on_pc is set to False. Defaults to "curvature".