
Weighted Point Cloud Normal Estimation
A weighted normal regression method for point clouds: a weight regressor predicts how relevant each neighbouring point is to the central point, and contrastive pre-training pairs each patch with its ground-truth normal, giving robust, feature-preserving normals on noisy synthetic and real-world point clouds.
Jul 1, 2023
Iteratively weighted principal component analysis and orientation consistency for normal estimation in point cloud
Iteratively Weighted PCA (IWPCA) estimates point cloud normals by repeating the local plane fit with bilateral spatial and normal weights, then makes the normal orientations consistent with a centroid-based pre-flip followed by minimal-spanning-tree propagation.
Jan 1, 2020