Reconstruction Algorithms for High-Energy Computed Tomography Images of Rocket Motors

ABSTRACT: High Energy X-ray computed tomography (CT) is used to inspect and evaluate the integrity of solid rocket motors (SRM"s). The driving purpose of this proposal is to improve the detection of defects in special cases where the defects are obscured due to limitations of the CT images. The expectation-maximization (EM) algorithm, which is a relatively new CT imaging algorithm, increases resolving power for showing defects more clearly. Lickenbrock Technologies has developed a variant of the EM algorithm and other software algorithms for improving fidelity of CT images. The limitation is long computation times with large image volumes involving 100's of CT slices. An approach to accelerating the computation speed significantly involves the utilization of many parallel graphical processing units (GPU"s) and many parallel core central processing units (CPU"s) on one computer system. Phase I will involve prototyping test versions of the EM and the other related algorithms, measuring processing times on representative scaled-down computer architectures, investigating the price and performance of a variety of high-end computer architectures that are commercially available and projecting processing times on each of these architectures based upon the measured times and relevant formulas for translating these time projections between computer architectures. BENEFIT: The benefit to the Department of Defense will be the capability to deploy improved computed tomography (CT) algorithms for improving defect detection in solid rocket motor inspection. Commercial and other inspection applications, involving CT, that may benefit include automotive components, baggage handling at airports, and micro CT for small animal and small materials imaging.