Abstract

A new design method, the maximized amplitude difference (MAD) method, is developed to improve the discrimination of binary phase-only filters (BPOFs). In this approach, we construct an energy function that represents the weighted difference between the amplitude of the autocorrelation peak caused by the target signal and that of the cross-correlation peak caused by the nontarget signal. The weight determines the suppression level of the crosscorrelation peak. The best phases for each of the two peaks are determined through a fast deterministic search algorithm. A BPOF that minimizes the energy function is encoded by assigning each filter pixel according to the projections of the spectral components of the target and the nontarget signals onto the phasors representing the two peaks. For the examples used in the numerical studies, a 7.0 dB increase of the ratio of the autocorrelation peak intensity to the cross-correlation peak intensity has been achieved. For a given suppression level of the cross-correlation peak, the MAD method also maximizes the intensity difference between the two peaks. The MAD method is also compared with the simulated annealing method through numerical studies. The two methods generate almost identical filters, but the MAD method is much faster.

© 1992 Optical Society of America