Abstract

Digital computing operations are generally performed either by using Boolean logic gates or by truth table lookup processing. The latter method is very fast since the output is realized in one step but requires a large content-addressable memory.¹ Arithmetic processing may also be achieved by using optical pattern recognition systems. However, filter-based pattern recognition is not suitable for real-time operation since a complex filter is needed for each new input. A joint transform correlator (JTC) is particularly suitable for real-time optical pattern recognition since it does not involve filter fabrication. Herein, we investigate the application of a JTC for real-time optical arithmetic processing. As an illustration, we consider the implementation of a binary full adder with a JTC. But as the number of objects in the input joint image increases, the correlation signal is expected to suffer from reduced discrimination, large correlation sidelobes, large correlation bandwidth, and low optical efficiency, thus making the detection rather difficult. To alleviate these difficulties, therefore, we propose a multichannel JTC for implementing multi-input arithmetic operations. The performance of the proposed technique is finally verified by computer simulation.

© 1992 Optical Society of America