Performance of maximum entropy algorithms for image reconstruction from projections

Abstract
The analysis of maximum entropy algorithms is provided for image reconstruction from projections. The performance of block- and row-type multiplicative algebraic reconstruction techniques, Bregman's method of convex programming for entropy maximization under linear equality constraints, and MENT is compared with those of the Algebraic Reconstruction Techniques and the convolution backprojection. The quality of the successive reconstructed iterates versus the computational complexity of the algorithms is given. Furthermore, it is shown that a priori knowledge of the shape of the image to be reconstructed is easily included with these algorithms by solving a more general optimization problem.

This publication has 5 references indexed in Scilit: