Title: Provably Good Moving Least Squares
Speaker: Jin-Xin Huang

HKUST

Date: Friday Oct 28, 2005
Time 11-12 Venue: Room 3464, HKUST   


Abstract 
In this talk, I’ll talk about the analysis of Kolluri's  moving least squares algorithm for reconstructing a surface from point cloud data.

Implicit function I is defined and its zero set U is the reconstructed surface. It has been proved that I is a good approximation to the signed distance function of the sampled surface F and that U is geometrically close to and homeomorphic to F.
The proof requires sampling conditions similar to $epsilon$-sampling, used in Delaunay reconstruction algorithms.