# DAAD PPPI DST 2019 - High Fidelity Surface Reconstruction from Noisy Point Sets

The process of surface reconstruction aims at the generation of digital geometry models from real world shapes. 3D shape acquisition technologies such as laser scanners or imaging devices produce raw digital data sets (often point sets) of physical shapes which then need to be converted into high-quality geometric representations, e.g. triangle meshes or splines, to best replicate the physical shape. Inevitable problems in surface reconstruction is the presence of noise and outliers due to systemic deficiencies in the data acquisition. Also, surface reconstruction algorithms are often time-consuming due to complexity of the reconstruction algorithm.

In this project, we address key problems in surface reconstruction by developing a feature preserving point set denoising method as a preprocessing step to obtain a cleaned point set without losing feature information for high fidelity surface reconstruction. Furthermore, we are developing a local Delaunay based surface reconstruction algorithm, which takes the preprocessed point set as the input and produces a high-quality triangulated mesh with topological guarantee.

The key objective of this project is to remove spurious noise and outliers and to reconstruct a high-fidelity manifold surface with minimal running time algorithm complexity.

The project is a cooperation between the Geometry Processing group at FU Berlin and Advanced Geometric Computing Lab at IITM in Chennai support by DAAD and DST (Department of Science and Technology).