open3d.pipelines.registration.NormalDistributionsTransformOption#

class open3d.pipelines.registration.NormalDistributionsTransformOption#

Options for 3D Normal Distributions Transform registration.

__init__(*args, **kwargs)#

Overloaded function.

  1. __init__(self: open3d.pybind.pipelines.registration.NormalDistributionsTransformOption, arg0: open3d.pybind.pipelines.registration.NormalDistributionsTransformOption) -> None

Copy constructor

  1. __init__(self: open3d.pybind.pipelines.registration.NormalDistributionsTransformOption, voxel_size: typing.SupportsFloat = 1.0, min_points_per_voxel: typing.SupportsInt = 6, covariance_regularization: typing.SupportsFloat = 0.001, transformation_epsilon: typing.SupportsFloat = 1e-06, relative_objective: typing.SupportsFloat = 1e-06, max_iteration: typing.SupportsInt = 30, outlier_threshold: typing.SupportsFloat = 9.0, neighbor_search_type: typing.SupportsInt = 1) -> None

property covariance_regularization#

Minimum eigenvalue ratio used to regularize voxel covariances.

property max_iteration#

Maximum number of Gauss-Newton iterations.

property min_points_per_voxel#

Minimum number of target points needed for a voxel Gaussian.

property neighbor_search_type#

0 uses the rounded center voxel; 1 also uses the six face-adjacent voxels.

property outlier_threshold#

Maximum squared Mahalanobis distance accepted for a point-to-voxel residual.

property relative_objective#

Stop optimization when the relative change in mean Mahalanobis objective is lower than this value.

property transformation_epsilon#

Stop optimization when the update vector norm is lower than this value.

property voxel_size#

Target voxel size used to build the Gaussian model.