I have a large number of points representing several geological beds with many outliers. I have tried Ransac, DBSCAN, GMM clustering to differentiate between the point clusters representing different beds but they fail in complex scenarios e.g. when two beds are crossing. I have used Open3d, but there I can manually separate the points only.
Is there any way I can extract possible surfaces (representative of beds) from the point cloud automatically?
Any help in this direction will be much appreciated. Thanks