Robust Moving Least-squares Fitting with Sharp Features

1. Briefly summarize the paper’s contributions. Does it address a new problem? Does it present a new approach? Does it show new types of results?  [AS] This paper presents a new robust moving least squares (MLS) approach to define a surface from a potentially noisy point set. Using techniques from robust statistics to determine neighborhoods used by MLS, this approach produces a piecewise smooth surface, and is therefore able to preserve sharp features.  [DS] The paper introduces a robust moving least-squares technique for reconstructing a piecewise smooth surface from a noisy point cloud. The method introduces the use of a new robust statistics method for outlier detection: the forward-search paradigm. The algorithm classifies regions of a point-set into outlier-free smooth regions, which allows for the projection of points onto a locally smooth region. This increases the stability of the projection. The algorithm synthesizes new points that reconstruct sharp crease features, which are not part of the input set.  [FP] The authors propose a method to do point projection on a piecewise smooth surface. The authors decompose the set of samples in several regions using robust statistical methods (Least Median Squares and Forward Search) and project the point to the closest valid region. The results presented by the authors show reconstructions with sharper edges than global MLS (e.g., PSS) and also seems to be more robust to noise and outliers.  [JD] The paper presents a method for point set surfaces that detects sharp features. Previous work smoothed sharp features away with noisy samples. This work allows a surface with sharp edges to be reconstructed from a noisy input point set.  [LF] The paper introduces the forward-search paradigm, a robust statistics method for outlier detection, as a technique for dealing with noise and retaining sharp features when computing an MLS-surface.  [MK] The paper proposes two new contributions to the area of surface reconstruction. First, it describes a robust forward-search paradigm to generate reconstructions that are stable in the presence of noise. Second, it extends the MLS framework by supporting the reconstruction of surfaces with sharp features.