ABSTRACT
Optical pattern recognition is a well-known research field since the pioneer work of VanderLugt [1]. Several books and review articles survey different approaches of this field [2-5]. The optical correlator is the main component in many of the optical image recognition schemes. Spatial correlations can be done fast and in parallel by the use of optics. The functions involved in such correlators are at most two dimensional (2D). However, our real spatial world is three dimensional (3D), and in some applications, 3D objects should be processed along all of their three dimensions. In general, 2D pattern recognition systems cannot determine the exact longitudinal distances between the various targets and cannot map the identified targets in the 3D space. In other words, with a 2D correlator, we cannot be sure which object is in front or behind other objects. For that purpose, we need to extend the correlation from two dimensions to three. The 3D correlation has two advantages over the conventional 2D correlation. First, we employ the information obtained from the 3D shape of the object, including its pattern along its depth dimension. Second, the target’s location in the 3D space is exactly identified by the correlator. This essential information might be useful for pattern recognition, image reconstruction, and target tracking systems.
