open3d.pipelines.registration.registration_ransac_based_on_feature_matching#

open3d.pipelines.registration.registration_ransac_based_on_feature_matching(source: open3d.geometry.PointCloud, target: open3d.geometry.PointCloud, source_feature: open3d.pipelines.registration.Feature, target_feature: open3d.pipelines.registration.Feature, mutual_filter: bool, max_correspondence_distance: SupportsFloat, estimation_method: open3d.pipelines.registration.TransformationEstimation = TransformationEstimationPointToPoint(with_scaling=False), ransac_n: SupportsInt = 3, checkers: collections.abc.Sequence[open3d.pipelines.registration.CorrespondenceChecker] = [], criteria: open3d.pipelines.registration.RANSACConvergenceCriteria = RANSACConvergenceCriteria(max_iteration=100000, confidence=9.990000e-01)) → open3d.pipelines.registration.RegistrationResult#

Function for global RANSAC registration based on feature matching

Parameters:

  • source (open3d.geometry.PointCloud) – The source point cloud.

  • target (open3d.geometry.PointCloud) – The target point cloud.

  • source_feature (open3d.pipelines.registration.Feature) – Source point cloud feature.

  • target_feature (open3d.pipelines.registration.Feature) – Target point cloud feature.

  • mutual_filter (bool) – Enables mutual filter such that the correspondence of the source point’s correspondence is itself.

  • max_correspondence_distance (SupportsFloat) – Maximum correspondence points-pair distance.

  • estimation_method (open3d.pipelines.registration.TransformationEstimation, optional, default=TransformationEstimationPointToPoint(with_scaling=False)) – Estimation method. One of (TransformationEstimationPointToPoint, TransformationEstimationPointToPlane, TransformationEstimationForGeneralizedICP, TransformationEstimationForColoredICP)

  • ransac_n (SupportsInt, optional, default=3) – Fit ransac with ransac_n correspondences

  • checkers (collections.abc.Sequence[open3d.pipelines.registration.CorrespondenceChecker], optional, default=[]) – Vector of Checker class to check if two point clouds can be aligned. One of (CorrespondenceCheckerBasedOnEdgeLength, CorrespondenceCheckerBasedOnDistance, CorrespondenceCheckerBasedOnNormal)

  • criteria (open3d.pipelines.registration.RANSACConvergenceCriteria, optional, default=RANSACConvergenceCriteria(max_iteration=100000, confidence=9.990000e-01)) – Convergence criteria

Returns:

open3d.pipelines.registration.RegistrationResult