NDTnet 1998 Aug, Vol.3 No.8

Abstract

Image reconstruction from projections is a basic process for computerized tomography (CT). The CT performance strongly depends on the object and noise properties. Modem CT theory is based on taking into account prior knowledge about these properties but they are usually undefined. One possible way to avoid these difficulties consist in employing principles of adaptive processing [1]. The basis of the adaptive computerized tomography (ACT) is the estimation of the object and noise properties during the current measurement of the projections. The aim of this work is to develop the theory and the construction principles for the ACT. For this purpose some results of automatic control and identification theory are employed. The tomographic process is represented like an identification system with adaptive tuned parameters of the object model. These parameters represent the material properties of the examined object. The theory of ACT is based on employing the stochastic approximation methodology and consists of recursive probability algorithms. The initial convergence properties of these algorithms weakly depend on the introduced prior knowledge. During the reconstruction process the information about the object and the noise properties is accumulated which influences the convergence of the algorithm. Therefore the image reconstruction process employs a posteriori information about the specific object and noise properties instead of using a priori knowledge. This specific information is more useful in comparison with statistical prior knowledge about a big set of objects. The main problem of this theory is to obtain an appropriate convergence criteria and an estimation of the reconstruction performance due to the number of projections. The construction principles include an adaptive measurement channel for the estimation of the object and noise properties. The results of this investigation are to obtain suitable reconstruction algorithms, and the estimation of their performance and their convergence velocity.

1. Tsypkin Y. Z, Information theory for identification. Moscow, Wauka, 1984, pp. 320.

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