Bereiche | Tage | Auswahl | Suche | Downloads | Hilfe
M: Metallphysik
M 34: Symposium Tomographic Methods in Materials Research
M 34.1: Vortrag
Montag, 7. März 2005, 17:00–17:20, TU H1058
Application of discrete tomography to multi-level images and real projection data — •Zoltán Kiss1, Attila Kuba1, Antal Nagy1, and Márton Balaskó2 — 1Dept. of Image Processing and Computer Graphics, University of Szeged, Hungary, H-6701 Szeged, Hungary, P. O. Box 652. — 2KFKI Atomic Energy Research Institute, H-1525 Budapest 114, Hungary, P.O. Box 49.
Discrete tomography is an imaging technique to reconstruct discrete images from their projections using the knowledge that the object to be reconstructed contains only a few homogeneous materials characterized by known discrete absorption values. By this assumption the reconstruction can be done from relatively few projections and efficiently used in industrial non-destructive material examination. In previous research a stochastic reconstruction technique was successfully applied to binary phantom images, which considers the reconstruction as an optimization task. Now an extension of this method is introduced, which reconstructs multi-level phantom images from a few noisy projections and additionally we discuss the effect of several reconstruction parameters. We provide our experiments based on real measured projections including the possible improvements, reduction of the noise effect, building in a priori information, and a comparison with the classical FBP reconstruction method.