Occluder Frequency Analysis for Evaluating the Level of Visibility of Partly Occluded Objects

Information

• Publication Type: Bachelor Thesis
• Workgroup(s)/Project(s):
  • Smart Communities and Technologies: 3D Spatialization
• Date: July 2022
• Date (Start): 1. January 2022
• Date (End): 15. July 2022
• Matrikelnummer: 01518693
• First Supervisor:
  • Michael Wimmer
  • Bernhard Kerbl
• Keywords: rendering, occlusion culling, real-time

Abstract

To increase rendering efficiency of large and complex scenes, occlusion culling algorithms detect objects which are completely hidden by others and therefore do not need to be rendered. However, these methods often follow an all-or-nothing principle, either culling the geometry entirely or drawing it at full detail. This approach disregards an important subcategory of the visibility problem: detecting objects that are hardly visible because they are partly occluded and which can therefore be rendered at a lower level of detail without generating noticeable artifacts. In this thesis we assess the level of visibility of such objects by computing a hierarchical occlusion map and analysing its structure based on the frequencies of the occluders. This analysis results in a parameter that controls the level of detail (LOD) in which the geometry is rendered. The algorithm performs well even in scenes with sparse occlusion, surpassing the standard hierarchical occlusion map algorithm, with still a lot of potential for even further improvements.

Additional Files and Images

Weblinks

No further information available.

BibTeX

@bachelorsthesis{kristmann-2022-occ,
  title =      "Occluder Frequency Analysis for Evaluating the Level of
               Visibility of Partly Occluded Objects",
  author =     "Elias Kristmann",
  year =       "2022",
  abstract =   "To increase rendering efficiency of large and complex
               scenes, occlusion culling algorithms detect objects which
               are completely hidden by others and therefore do not need to
               be rendered. However, these methods often follow an
               all-or-nothing principle, either culling the geometry
               entirely or drawing it at full detail. This approach
               disregards an important subcategory of the visibility
               problem: detecting objects that are hardly visible because
               they are partly occluded and which can therefore be rendered
               at a lower level of detail without generating noticeable
               artifacts. In this thesis we assess the level of visibility
               of such objects by computing a hierarchical occlusion map
               and analysing its structure based on the frequencies of the
               occluders. This analysis results in a parameter that
               controls the level of detail (LOD) in which the geometry is
               rendered. The algorithm performs well even in scenes with
               sparse occlusion, surpassing the standard hierarchical
               occlusion map algorithm, with still a lot of potential for
               even further improvements.",
  month =      jul,
  address =    "Favoritenstrasse 9-11/E193-02, A-1040 Vienna, Austria",
  school =     "Research Unit of Computer Graphics, Institute of Visual
               Computing and Human-Centered Technology, Faculty of
               Informatics, TU Wien ",
  keywords =   "rendering, occlusion culling, real-time",
  URL =        "https://www.cg.tuwien.ac.at/research/publications/2022/kristmann-2022-occ/",
}