Information

  • Publication Type: Bachelor Thesis
  • Workgroup(s)/Project(s):
  • Date: September 2022
  • Date (Start): 4. February 2022
  • Date (End): 4. September 2022
  • Matrikelnummer: 01635282
  • First Supervisor: Eduard GröllerORCID iD

Abstract

The recently achieved differentiability of path-tracing algorithms, which are the standard for generating photo-realistic images, opens up optimization possibilities for the 3Dreconstruction of an object using images acquired by X-ray scans. The reconstruction is accomplished by ”inverting the rendering pipeline”, which in practice means obtaining 3D scene parameters, such as volumetric data, from 2D images. The images act as a reference in our algorithm, which is optimizing the scene parameters until the image acquired by our rendered scene minimally differs from the reference image. In this publication, we represent a proof-of-concept and early experiments for differential rendering for CT reconstruction. Our implementation is able to successfully reconstruct the geometry and volume of specimens, using only images acquired from a software-simulated X-ray scan.

Additional Files and Images

Additional images and videos

Additional files

Weblinks

No further information available.

BibTeX

@bachelorsthesis{Vucenovic_2022,
  title =      "Differential Rendering for Computed Tomography
               Reconstruction",
  author =     "Aleksandar Vucenovic",
  year =       "2022",
  abstract =   "The recently achieved differentiability of path-tracing
               algorithms, which are the standard for generating
               photo-realistic images, opens up optimization possibilities
               for the 3Dreconstruction of an object using images acquired
               by X-ray scans. The reconstruction is accomplished by
               ”inverting the rendering pipeline”, which in practice
               means obtaining 3D scene parameters, such as volumetric
               data, from 2D images. The images act as a reference in our
               algorithm, which is optimizing the scene parameters until
               the image acquired by our rendered scene minimally differs
               from the reference image. In this publication, we represent
               a proof-of-concept and early experiments for differential
               rendering for CT reconstruction. Our implementation is able
               to successfully reconstruct the geometry and volume of
               specimens, using only images acquired from a
               software-simulated X-ray scan. ",
  month =      sep,
  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 ",
  URL =        "https://www.cg.tuwien.ac.at/research/publications/2022/Vucenovic_2022/",
}