skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Leveraging GPUs in Industrial Computed Tomography Reconstruction.


Abstract not provided.

; ;
Publication Date:
Research Org.:
Sandia National Lab. (SNL-NM), Albuquerque, NM (United States)
Sponsoring Org.:
USDOE National Nuclear Security Administration (NNSA)
OSTI Identifier:
Report Number(s):
DOE Contract Number:
Resource Type:
Resource Relation:
Conference: Proposed for presentation at the Computer Science Seminar Presentation at the University of Washingotn in Seattle, WA.
Country of Publication:
United States

Citation Formats

Orr, Laurel Jeffers, Jimenez, Edward Steven,, and Thompson, Kyle R. Leveraging GPUs in Industrial Computed Tomography Reconstruction.. United States: N. p., 2015. Web.
Orr, Laurel Jeffers, Jimenez, Edward Steven,, & Thompson, Kyle R. Leveraging GPUs in Industrial Computed Tomography Reconstruction.. United States.
Orr, Laurel Jeffers, Jimenez, Edward Steven,, and Thompson, Kyle R. 2015. "Leveraging GPUs in Industrial Computed Tomography Reconstruction.". United States. doi:.
title = {Leveraging GPUs in Industrial Computed Tomography Reconstruction.},
author = {Orr, Laurel Jeffers and Jimenez, Edward Steven, and Thompson, Kyle R.},
abstractNote = {Abstract not provided.},
doi = {},
journal = {},
number = ,
volume = ,
place = {United States},
year = 2015,
month = 1

Other availability
Please see Document Availability for additional information on obtaining the full-text document. Library patrons may search WorldCat to identify libraries that hold this conference proceeding.

Save / Share:
  • Modern high resolution medical computed tomography (CT) scanners can produce geometrically accurate sectional images of many types of industrial objects. Computer software has been developed to convert serial CT scans into a 3-D surface form, suitable for display on the scanner itself. This software, originally developed for imaging the skull, has been adapted for application to industrial CT scanning, where serial CT scans through an object of interest may be reconstructed to demonstrate spatial relationships in three dimensions that cannot be easily understood using the original slices. The methods of 3-D reconstruction and solid modeling are reviewed, and reconstruction inmore » three dimensions from CT scans through familiar objects is demonstrated.« less
  • Domain-iterative algorithms-which iterate between projection, Radon, Fourier, and spatial domains for extrapolation of projections-have been shown to give improved limited-data computed tomography reconstructions. These methods generally have one or more interpolation steps in the iterations that degrade the image resolution at the same time that they attempt to improve the extrapolation. Therefore, it is important to converge quickly to reduce both the computation time and the number of interpolation steps. We describe a class of hybrid domain-iterative algorithms for 2D image reconstruction. A particular algorithm, called the consistent iterative reconstruction-reprojection (CIRR) algorithm, was designed and implemented along with other classicalmore » domain-iterative methods. The algorithms are tested the hollow projections problem, where simple object models are assumed known. Results show that the CIRR algorithm is robust, accurate, and converges quickly.« less
  • In support of stockpile stewardship and other important missions, Los Alamos is continually looking for fast and effective ways of inspecting and evaluating industrial parts. Thus, Los Alamos is continually striving to improve our radiography and computed tomography (CT) capabilities. Cormack and Hounsfield received the Nobel Prize in 1979 for their pioneering work in computed tomography that led to the development of medical scanners. Copley et al. provides a good history of the development of industrial CT systems. The early systems collect data via a single detector or linear detector array. While CT offers greatly increased spatial resolutions over radiography,more » CT inspections with a linear array are slow and costly. To improve the viability of CT for NDT applications, Feldkamp, Davis, and Kress reported a cone beam reconstruction technique that speeds up the CT process by using image data rather than data collected by a linear array. Because it potentially offers processing speeds up to 10 times faster than CT systems that use a linear array, we are building a cone beam CT for use with our 20 MV x-ray source and Los Alamos Neutron Science Center (LANSCE) neutron sources. Our software, called CoBRA, is a portable cone beam reconstruction code for CT applications that efficiently and rapidly reconstructs large data sets. CoBRA applications include both x-ray and neutron inspections using x-ray phosphor screens coupled to either a CCD camera or flat-panel amorphous silicon arrays. Photographs of two amorphous silicon arrays.« less
  • Abstract not provided.
  • Abstract not provided.