Multi-slice tomographic reconstruction: to couple or not to couple

Preeti Gopal, Sharat Chandran, Imants Svalbe, Ajit Rajwade

Research output: Chapter in Book/Report/Conference proceedingConference PaperResearchpeer-review

2 Citations (Scopus)


Recent work in tomography focuses on algorithms that enable faster and more accurate reconstruction from as few measurements as possible. We review the advantage of jointly reconstructing multiple slices and show that joint reconstruction may suffer in the presence of adjacent dissimilar slices. This gives rise to the need to detect similarity or dissimilarity of unknown images before performing joint reconstruction. We propose a method to detect 'similar' slices directly from their tomographic measurements and juxtapose these similar slices. Since the images themselves are not available by definition, we compute similarity between slices based on image moments; these in turn are estimated in a novel way from Radon projection moments. A segmented least squares algorithm is then designed to couple only similar slices. Our results confirm the benefit of this method for tomographic reconstruction.

Original languageEnglish
Title of host publicationProceedings
Subtitle of host publicationTenth Indian Conference on Computer Vision, Graphics and Image Processing
EditorsDhruv Batra, Vijay Natarajan, Michael Brown
Place of PublicationNew York NY USA
PublisherAssociation for Computing Machinery (ACM)
Number of pages7
ISBN (Electronic)9781450347532
Publication statusPublished - 18 Dec 2016
EventIndian Conference on Computer Vision, Graphics and Image Processing 2016 - Indian Institute of Technology Guwahati, Assam, India
Duration: 18 Dec 201622 Dec 2016
Conference number: 10th (front matter)


ConferenceIndian Conference on Computer Vision, Graphics and Image Processing 2016
Abbreviated titleICVGIP 2016
Internet address


  • Compressive sensing
  • Image moments
  • Radon projection moments
  • Tomographic reconstruction

Cite this