• DocumentCode
    3358636
  • Title

    Aligning tilt slices for 3D TEM tomography based on 2D to 1D Radon transform

  • Author

    Chen, Wei-Jun

  • Author_Institution
    Software Dev., Carl Zeiss Meditec AG, Jena, Germany
  • fYear
    2010
  • fDate
    26-29 Sept. 2010
  • Firstpage
    609
  • Lastpage
    612
  • Abstract
    Aligning a tilt series in 3D TEM tomography is not only a shift compensation problem among tilt slices, but rather a parameter registration problem, in which a tilt axis should be registered on each slice with its position/orientation. A new method for 3D tilt alignment is suggested in this paper based on analyzing 1D signals resulting from projecting individual 2D slices in various directions (2D Radon transform). The axis orientation and position are identified by techniques of 1D shift compensation, 2D frequency analysis, etc. Without any information loss during the calculating process, the accuracy and the robustness for axis registration are ensured. Meanwhile, essentially working on 1D data provides the concrete base of computing performance. Based on this method, no any extra work of adding fiducial markers into samples, or detecting corners/edges from slice images are required; therefore both experimental and computational efforts could be heavily reduced.
  • Keywords
    Radon transforms; image registration; tomography; transmission electron microscopy; 2D frequency analysis; 3D TEM tomography; Radon transform; parameter registration; shift compensation problem; tilt slice alignment; Image reconstruction; Three dimensional displays; Tomography; Transforms; Transmission electron microscopy; Cross-correlation; Fractional Fast Fourier Transform; Radon transform; TEM tomography;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Image Processing (ICIP), 2010 17th IEEE International Conference on
  • Conference_Location
    Hong Kong
  • ISSN
    1522-4880
  • Print_ISBN
    978-1-4244-7992-4
  • Electronic_ISBN
    1522-4880
  • Type

    conf

  • DOI
    10.1109/ICIP.2010.5653010
  • Filename
    5653010