Binarization of a super‐resolving graytone pupil filter by digital halftoning

Six digital‐halftoning procedures, including one algorithm proposed by us, are compared to determine which one is best suited to binarization of a parabolic super‐resolving pupil filter. The procedures we deal with include iterative, error‐diffusion, error‐convergence, and 1‐pixel algorithms. We carry out a numerically simulated experiment in which an object that consists of either one point source or two coherent point sources is imaged in a 4f imaging system with either a continuous super‐resolving parabolic filter or one of its six different binary versions. The performance of binary filters is examined in terms of two parameters: the resemblance of their amplitude impulse response (AIR) to the AIR of the original continuous filter, as well as the two‐point Sparrow resolution criterion. We found that the best performance in terms of both figures of merit is achieved with the filter generated by means of one‐weight error diffusion when the weight is randomly positioned and the algorithm is processed on a serpentine raster.