Information

  • Publication Type: Bachelor Thesis
  • Workgroup(s)/Project(s):
  • Date: January 2022
  • Date (Start): 23. March 2020
  • Date (End): 24. January 2022
  • Matrikelnummer: 01426150
  • First Supervisor: Stefan OhrhallingerORCID iD
  • Keywords: change detection, uncertainty, 3d scanning

Abstract

The automated detection of changes in a 3D space can be a useful tool. [PCBS16] names 3D surface reconstruction, environment monitoring, natural events management, and forensic science as possible application scenarios. In this work, we introduce software that scans an area at two different points in time and detects the changes between these scans. The software is based on InfiniTAM [PKG+17], a framework released under an Oxford University License. InfiniTAM integrates multiple depth images (e.g. recorded with a Kinect-V2-Camera) to a 3D model using volumetric representations. Because of the volumetric representation and the fast GPU computation, the change detection can happen in real-time. This is outstanding because in other approaches, (like [PCBS16]) the change detection can take minutes. Other approaches that detect changes in real-time (like [KMK+19]) use the same representation of data (T-SDF) as we do. Our approach also takes sensor tolerance into account, which leads to a reduction of false change detections. This work can be seen as a starting point for more specific use cases (like [LTW+21]) who specify on scene change detection and overcoming unnecessary changes such as light and seasons.

Additional Files and Images

Additional images and videos

Additional files

Weblinks

No further information available.

BibTeX

@bachelorsthesis{steinschauer-2020-baa,
  title =      "Change Detection using the InfiniTAM framework",
  author =     "Thomas Steinschauer",
  year =       "2022",
  abstract =   "The automated detection of changes in a 3D space can be a
               useful tool. [PCBS16] names 3D surface reconstruction,
               environment monitoring, natural events management, and
               forensic science as possible application scenarios. In this
               work, we introduce software that scans an area at two
               different points in time and detects the changes between
               these scans. The software is based on InfiniTAM [PKG+17], a
               framework released under an Oxford University License.
               InfiniTAM integrates multiple depth images (e.g. recorded
               with a Kinect-V2-Camera) to a 3D model using volumetric
               representations. Because of the volumetric representation
               and the fast GPU computation, the change detection can
               happen in real-time. This is outstanding because in other
               approaches, (like [PCBS16]) the change detection can take
               minutes. Other approaches that detect changes in real-time
               (like [KMK+19]) use the same representation of data (T-SDF)
               as we do. Our approach also takes sensor tolerance into
               account, which leads to a reduction of false change
               detections. This work can be seen as a starting point for
               more specific use cases (like [LTW+21]) who specify on scene
               change detection and overcoming unnecessary changes such as
               light and seasons.",
  month =      jan,
  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 =   "change detection, uncertainty, 3d scanning",
  URL =        "https://www.cg.tuwien.ac.at/research/publications/2022/steinschauer-2020-baa/",
}