Kinect Fusion - Reconstruction

Information

• Publication Type: Bachelor Thesis
• Workgroup(s)/Project(s):
  • Reciprocal Shading for Mixed Reality
• Date: February 2013
• Date (Start): 1. October 2011
• Matrikelnummer: 0826080
• First Supervisor:
  • Michael Wimmer
  • Martin Knecht
• Keywords: Reconstruction, Kinect Fusion

Abstract

The procedure of collecting 3D data via an input device and processing it to a virtual 3D model is called 3D reconstruction. It is a widely used technique in visual computing, since modern applications like games or visualizations tend to be more and more photo-realistic leading to high costs in content creation. By using 3D reconstruction high quality geometry can be generated out of real objects. However to obtain good reconstructions special hardware is needed which is very expensive. Since Microsoft released the Kinect camera, which has a depth sensor in addition to the RGB-sensor, a quite cheap hardware is available that is able to extract 3D data of its surroundings. KinectFusion also developed by Microsoft is a technique that uses the Kinect camera for 3D reconstruction in real-time. In order to achieve real-time speed the algorithm is executed almost exclusively on the graphics card. Each frame the algorithm first gathers the information from the Kinect and processes it. After that it measures the camera’s position in space and fills a 3D volume with surface data. Finally a raycasting algorithm is used to extract isosurfaces out of the volume. During the work on the thesis we implemented the KinectFusion algorithm inside the RESHADE framework. The results and the implementation itself are presented as part of the thesis.

Additional Files and Images

Weblinks

No further information available.

BibTeX

@bachelorsthesis{jahrmann_klemens_KFR,
  title =      "Kinect Fusion - Reconstruction",
  author =     "Klemens Jahrmann",
  year =       "2013",
  abstract =   "The procedure of collecting 3D data via an input device and
               processing it to a virtual 3D model is called 3D
               reconstruction. It is a widely used technique in visual
               computing, since modern applications like games or
               visualizations tend to be more and more photo-realistic
               leading to high costs in content creation. By using 3D
               reconstruction high quality geometry can be generated out of
               real objects. However to obtain good reconstructions special
               hardware is needed which is very expensive. Since Microsoft
               released the Kinect camera, which has a depth sensor in
               addition to the RGB-sensor, a quite cheap hardware is
               available that is able to extract 3D data of its
               surroundings. KinectFusion also developed by Microsoft is a
               technique that uses the Kinect camera for 3D reconstruction
               in real-time. In order to achieve real-time speed the
               algorithm is executed almost exclusively on the graphics
               card. Each frame the algorithm first gathers the information
               from the Kinect and processes it. After that it measures the
               camera’s position in space and fills a 3D volume with
               surface data. Finally a raycasting algorithm is used to
               extract isosurfaces out of the volume. During the work on
               the thesis we implemented the KinectFusion algorithm inside
               the RESHADE framework. The results and the implementation
               itself are presented as part of the thesis.",
  month =      feb,
  address =    "Favoritenstrasse 9-11/E193-02, A-1040 Vienna, Austria",
  school =     "Institute of Computer Graphics and Algorithms, Vienna
               University of Technology ",
  keywords =   "Reconstruction, Kinect Fusion",
  URL =        "https://www.cg.tuwien.ac.at/research/publications/2013/jahrmann_klemens_KFR/",
}