Generalization of the Jared-Ennis method to complex transmittance objects for the generation of Synthetic Discriminant Function Filters

摘要

In this work we present a generalization to complex transmittance objects of the Jared-Ennis algorithm for the generation of Synthetic Discriminant Function filters (SDFs). The original algorithm consists of the resolution of a nonlinear system of equations by means of an iterative procedure, including a phase adaptation of the filter. The method shown here takes into account the modulation of liquid crystal displays (LCD) both for scene and filter, generalizing the problem to the complex plane. Considering this new method gives a more realistic picture as the LCD modulation gives a complex distribution of the scenes instead of only real values as considered before. For instance, we use a high contrast configuration to display the scenes. Moreover, the addition of new parameters to the problem allows us to consider filters other than the phase-only one. In our case, we use a phase-mostly configuration to display the filter and the metric optimized is the maximum correlation intensity, as in the original method. Simulated results are presented for a two-class problem, as well as experimental results obtained in a VanderLugt correlator. The filters produce the desired correlation response in both cases.