Information

Abstract

In recent years the amount of acquisition methods for point clouds has been increasing consequently and it is getting more and more interesting for society. Even if it is possible to render point clouds directly, nowadays there exist many more algorithms which deal with triangle meshes than point clouds. For example 3D printer software requires watertight meshes as input. This makes automatic conversion of point sets to triangle meshes an important research topic. The aim of this Bachelor Thesis was to implement a plugin for Scanopy (a point cloud editing and rendering program) which can convert point clouds with hundreds of millions of samples in such a detailed degree that the data exceeds common main memory sizes. Therefore, an out-of-core algorithm was needed. The used out-of-core Poisson surface reconstruction approach requires the sorting of the input point samples in a preprocessing step. In this Bachelor Thesis it is shown that the sorting of the data with an optimized multithreaded merge sort algorithm can improve the total required time for the reconstruction process significantly. Further, this work indicates a problem which occurs while reconstructing meshes with a Poisson based reconstruction approach from scans of an open terrain. The problem leads to large unnecessary triangles which hide the reconstructed surface. A very basic solution approach for this problem is also stated.

Additional Files and Images

Additional images and videos

Additional files

Weblinks

No further information available.

BibTeX

@bachelorsthesis{mazza-2012-bakk,
  title =      "Optimized Sorting for Out-of-Core Surface Reconstruction",
  author =     "Sebastian Mazza",
  year =       "2018",
  abstract =   "In recent years the amount of acquisition methods for point
               clouds has been increasing consequently and it is getting
               more and more interesting for society. Even if it is
               possible to render point clouds directly, nowadays there
               exist many more algorithms which deal with triangle meshes
               than point clouds. For example 3D printer software requires
               watertight meshes as input. This makes automatic conversion
               of point sets to triangle meshes an important research
               topic. The aim of this Bachelor Thesis was to implement a
               plugin for Scanopy (a point cloud editing and rendering
               program) which can convert point clouds with hundreds of
               millions of samples in such a detailed degree that the data
               exceeds common main memory sizes. Therefore, an out-of-core
               algorithm was needed. The used out-of-core Poisson surface
               reconstruction approach requires the sorting of the input
               point samples in a preprocessing step. In this Bachelor
               Thesis it is shown that the sorting of the data with an
               optimized multithreaded merge sort algorithm can improve the
               total required time for the reconstruction process
               significantly. Further, this work indicates a problem which
               occurs while reconstructing meshes with a Poisson based
               reconstruction approach from scans of an open terrain. The
               problem leads to large unnecessary triangles which hide the
               reconstructed surface. A very basic solution approach for
               this problem is also stated.",
  month =      may,
  address =    "Favoritenstrasse 9-11/E193-02, A-1040 Vienna, Austria",
  school =     "Institute of Computer Graphics and Algorithms, Vienna
               University of Technology ",
  keywords =   "surface reconstruction, out-of-core, point processing",
  URL =        "https://www.cg.tuwien.ac.at/research/publications/2018/mazza-2012-bakk/",
}