Reconstruction of Binary Images Represented on Equilateral Triangular Grid Using Evolutionary Algorithms

In this paper we study the problem of reconstructing a binary hexagonal image represented on an equilateral triangular grid from its projections. We propose an evolutionary algorithm for reconstructing this images from its projections in three directions. We will consider three projections as the points in the grid are triangles and by the symmetry of the grid there are three natural directions. Genetic operators, e.g., crossover and mutation operators are defined, as well as an operator based on the switch operator from discrete tomography. We use a compactness operator to eliminate the isolated points of an image and improve the quality of the image. The initial population and the children images are generated using the network flow algorithm with some restrictions due to the equilateral triangular grid and point shape and type. Some artificial binary images were used to test the efficiency of the algorithm.