MENT: A maximum entropy algorithm for reconstructing a source from projection data

摘要: Abstract This paper describes an iterative algorithm, MENT, which produces a maximum entropy solution to the problem of reconstructing source from discrete set projection data. Whereas MART algorithm considered by Lent in (“1976 Society Photographic Scientists and Engineers Conference Proceedings,” SPSE, Washington, D.C., 1977) uses rectangular grid represent source, MENT discretization is better suited problem. Unlike MART, does not require evaluation logarithms or exponentials. The storage requirements are also lower for MENT. Numerical examples given two-dimensional reconstruction five views. compared with regard convergence rates, artifact formation, stability against noise errors A three-dimensional version considered; we give example direct four