Vol. 31, Issue 4, pp. 675-689 (2001)

Abstract

The paper presents a concurrent hybrid pattern recognition system. The feature extractor of the system is based on optical properties of ring-wedge detector (RWD) or computer generated hologram (CGH) which serves as RWD. The classifier is made as artificial neural network (ANN). Since the feature extraction is an optical and thus fully concurrent process, hence such systems can be designed for real time pattern recognition if only the classifier of characteristic features works fast enough. In presented system the optimized by first author`s original method CGH is used instead of widely described standard one. A comparison of recognition results for both types of feature extractors is also provided. Finally, a methodology of obtaining fully concurrent system with optimized CGH and optical ANN playing the role of a classifier is proposed.