Skip to Main Content
The aim of this paper is to present a novel methodology for dynamic surface reconstruction from unorganized point cloud. The technique assumes to create an initial surface and its shape description basing on the training dataset. During the fitting process the initial surface evolves to describe the surface shape enclosed in the point cloud as exactly as possible. Additionally, on the evolving surface shape the constraints are imposed, which are responsible to keep the surface shape characteristics described in the training dataset. The steps of the methodology are described and the results of experiments are presented.