Robust point cloud segmentation and normal estimation

  • A new and more robust multivariate statistical method for point cloud normal and local curvature estimation

  • This method results in a more reliable estimation of normals in noisy 3D point clouds containing sharp features and planar transitions

  • A new region growing method for context-free segmentation of unstructured noisy point clouds with outliers

  • The segmentation method uses a locally adaptive spatial connectivity analysis to account geometric features during segment growth

  • Qualitative and quantitative evaluation of the presented method using a series of challenging point clouds

PCA - based Normal Estimation

Our Robust Normal Estimation Approach

Related Publications