<title>Multistage binary optical processing</title>

Abstract

Image processing in general and optical image processing in particular require very accurate and very complex processors. Such processors are sometimes difficult to manufacture and expensive to purchase. They also might be non flexible in their design. The principle of generating a single processor by use of several simpler processor-modules in cascade (and/or in parallel) is quite familiar. However, in optics this approach is mainly used for filtering in the Fourier or fractional Fourier planes. In this work the authors introduce multi-stage optical processing in the Fresnel plane. Using a small number of binary masks (either amplitude or phase) along the path of the light, one may process the incoming beam in the same manner as using a single high-resolution complex mask. The authors present an algorithm for establishing the binary processors and introduce the results obtained by this approach. An important application of this technique is the field of image recognition. Simulations demonstrate that minor manipulations on the input, affect the output plane significantly. On the other hand, hiding fractions of the input pattern hardly influence the output whereas the obtained effect reveals information regarding the flaw inserted within the input pattern.