NDTnet 1998 Aug, Vol.3 No.8

Abstract

From the theory of classic Computer Tomography (CT) it follows a strict limitations from below in the number of exposed projections and accessible views for the observation of the object under reconstruction. To overcome lack of data, caused by pointed limitations, we introduced in the iteration algorithm the a priori knowledge in the form of Gibbs priors using Random Markov Fields (RAE) having clique structure. We proposed the MRF with cliques of intermediate order to calculate the Gibbs energy, which is minimized during iteration and developed the algorithms and software for workstation and PC which makes it possible to 50-100 times reducing the number of needed projections (from 720 for conventional CT to 3-7) and to use very small angles for observation of the object (about 90° -100°).

Important progress was done in the estimation and appropriate selection for the initial conditions before starting the iteration process. This was called first-step reconstruction caused the multi-step concept as the available computer time consuming procedure, which provides the convergence of iteration process. The sense of primary steps is to reduce strongly the space of interest before minimization, which was called Virtual Defects Space (VDS).

We developed a special algorithm for digital simulation of the X-ray absorption in the voxels lattice. Objectively this is fine splitting of the object's space which is important to increase the space resolution. Its introducing causes the possibility of tomographic imaging of very thin defects like cracks.All this parts were combined in the reconstruction algorithms and software for workstation and PC for 3-D image restoration. They were checked using both simulated and real experimental data. The theory and results are partially described in some published and accepted papers [ 1 -2].

This report disseminates the new approach for the reconstruction of more complicated structures which a priory knowledge is less defined then for binary structures. The wide potentialities in this way are shown. Namely three phase structures like pores and inclusions of foreign material in matrix are considered. The developed algorithms are applied for the 3D reconstruction of real objects and its quality is compared with that provided by classic CT.

References:
1. V.L.Vengrinovich, Yu.B.Denkevich, G.-R.Tillack, C.Nockeman. In Review of progress in QNDE 16, v.A, eds. By D.O.Thompson and D.E.Chimenti, Plenum Press, N-Y, 1997, pp.317-323.
2. V.L.Vengrinovich, Yu.B.Denkevich, G.-R.Tillack, C.Nockeman. Reconstruction of 3-D Binary Structures from Extreme Limited Number of Cone-Beam X-ray Projections. Choice submitted to the Journ. IEEE Transactions on Nuclear Sciences, 1996.

Abstract Source:
Book of Abstracts, 7th European Conference on Non-Destructive Testing, 26-29 May 1998, ISBN: 87-986898-0-00

Full-Text Source:
Proceedings of the 7th European Conference on Non-Destructive Testing, 26-29 May 1998, ISBN:

Publication Contact:
7th ECNDT, Park Alle 345, DK-2605 Broendby, Denmark, Fax: +45 46 26 70 11, Email: 101373.3414@compuserve.com