Evolutionary reconstruction of gray images

A GA-based technique for reconstructing plane gray images from projections is presented. The algorithm estimates two-dimensional images from their four one-dimensional projection data. For generation of an initial population, an (iterative) algebraic reconstruction technique (ART) is applied. A Laplacian constraint term is included in the fitness function for handling smooth images. The proposed evolutionary method works well for improving poorly reconstructed ART images from a limited number of projection directions