1 February 1999 Experimental optical results with minimum Euclidean distance optimal filters, coupled modulators, and quadratic metrics
Author Affiliations +
Optical Engineering, 38(2), (1999). doi:10.1117/1.602089
Abstract
Optical correlators used in pattern recognition should have their operation optimized according to physically measurable behavior related to the correlator's task. The optimization is best done with explicit consideration of the limited set of complex values available to the filter. We earlier developed a theoretical filter computing strategy, MEDOF (minimum Euclidean distance optimal filter), that optimizes a wide variety of correlation metrics on physically realistic modulators. We use a single reference image and add noise according to a stated model. We show laboratory results for filters optimizing a variety of metrics on a highly coupled SLM. The results show improvements in the metrics by using MEDOF when compared with some other filter computing strategies such as matching phase. Optimizing SNR appears preferable to the other metrics when detecting an object in clutter of known power spectral density.
Richard D. Juday, R. Shane Barton, Stanley E. Monroe, "Experimental optical results with minimum Euclidean distance optimal filters, coupled modulators, and quadratic metrics," Optical Engineering 38(2), (1 February 1999). https://doi.org/10.1117/1.602089
JOURNAL ARTICLE
11 PAGES


SHARE
RELATED CONTENT

Selection of LCTV operating curves for input and filter
Proceedings of SPIE (July 01 1992)
Optimal filters: a unified approach for SNR and PCE
Proceedings of SPIE (January 12 1993)
How do we make a good input image?
Proceedings of SPIE (March 27 1997)

Back to Top