Achieving super-resolution with band-pass images

The paper addresses the problem of spectrum extrapolation for two-dimensional images of objects with compact support and Fourier transform known in a circular band-pass region. The problem is treated by exploiting commonly available a priori knowledge (reality and positivity) in order to define the subsets of the Hilbert space L?2 to be used with a Projection Onto Convex Sets (POCS) algorithm. This technique consists in orthogonally projecting the current estimated image onto a number of closed and convex subsets cyclically, and is shown to be a generalization of the well-known Gerchberg-Papoulis (GP) algorithm. Some experimental results, with images obtained via a backscattering-based electramagnetic tomographic technique, are also reported, showing that the algorithm is practical and very effective, especially in suppressing the strang ambiguity artifacts inherent in the particular type of images treated.