The application of parallel processing techniques in coded aperture imaging
The application of parallel processing techniques in coded aperture imaging
The availability of low-cost, high performance computer systems based on parallel architectures is revolutionising many scientific disciplines. In gamma-ray astronomy, the amount of data processing required to obtain useful astrophysical results from observational data has historically been large, and this has provided an early motivation for experimentation with parallelism.
This thesis reports the application of parallel processing techniques using transputer arrays to computational problems in gamma-ray astronomical imaging. The processing of images from unstable or poorly-stabilised coded aperture telescope platforms is first considered, with particular reference to the ZEBRA coded aperture telescope.
ZEBRA is a low energy (200 keV to 10 MeV) coded aperture gamma-ray telescope with an arcminute-level source location capability which is designed to operate at stratospheric altitudes. The stabilised platform which is to carry ZEBRA suffers from image blurring caused by inexact tracking of the target region, the effect of which is to considerably increase the computational requirements for image reconstruction. Parallelised image reconstruction and processing algorithms are developed to compensate for this blurring.
The fast Fourier transform is identified as an important computational image processing tool which requires an efficient parallel implementation on transputers. A highly-efficient parallel FFT algorithm is developed for a transputer network employing a hypercube architecture. The algorithm is shown to be scalable for large networks of transputers by using a dynamic reconfiguration strategy. The implementation of parallel FFTs on next-generation parallel computers with virtual complete connectivity is then considered.
University of Southampton
1992
Duncan, Stephen Howard
(1992)
The application of parallel processing techniques in coded aperture imaging.
University of Southampton, Doctoral Thesis.
Record type:
Thesis
(Doctoral)
Abstract
The availability of low-cost, high performance computer systems based on parallel architectures is revolutionising many scientific disciplines. In gamma-ray astronomy, the amount of data processing required to obtain useful astrophysical results from observational data has historically been large, and this has provided an early motivation for experimentation with parallelism.
This thesis reports the application of parallel processing techniques using transputer arrays to computational problems in gamma-ray astronomical imaging. The processing of images from unstable or poorly-stabilised coded aperture telescope platforms is first considered, with particular reference to the ZEBRA coded aperture telescope.
ZEBRA is a low energy (200 keV to 10 MeV) coded aperture gamma-ray telescope with an arcminute-level source location capability which is designed to operate at stratospheric altitudes. The stabilised platform which is to carry ZEBRA suffers from image blurring caused by inexact tracking of the target region, the effect of which is to considerably increase the computational requirements for image reconstruction. Parallelised image reconstruction and processing algorithms are developed to compensate for this blurring.
The fast Fourier transform is identified as an important computational image processing tool which requires an efficient parallel implementation on transputers. A highly-efficient parallel FFT algorithm is developed for a transputer network employing a hypercube architecture. The algorithm is shown to be scalable for large networks of transputers by using a dynamic reconfiguration strategy. The implementation of parallel FFTs on next-generation parallel computers with virtual complete connectivity is then considered.
This record has no associated files available for download.
More information
Published date: 1992
Identifiers
Local EPrints ID: 462266
URI: http://eprints.soton.ac.uk/id/eprint/462266
PURE UUID: 8dd4b315-53f5-4515-ad24-482151655a51
Catalogue record
Date deposited: 04 Jul 2022 19:04
Last modified: 04 Jul 2022 19:04
Export record
Contributors
Author:
Stephen Howard Duncan
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics