1 August 1996 Pattern classification using a joint transform correlator based nearest neighbor classifier
Author Affiliations +
Abstract
A JTC-based nearest neighbor classifier (JTC-NNC) is presented, by which shift-invariant pattern classification can be obtained. To efficiently utilize the spatial domain input plane, a non-zero-order JTC is introduced to remove the autocorrelation power spectra. In addition, a phase-transform technique is introduced into the JTC-NNC to improve the light efficiency and pattern discriminability. Finally, application of the JTC-NNC to optical character recognition is discussed, and computer simulation is provided to show the feasibility of the JTC-NNC.
Guowen Lu, Guowen Lu, Francis T. S. Yu, Francis T. S. Yu, } "Pattern classification using a joint transform correlator based nearest neighbor classifier," Optical Engineering 35(8), (1 August 1996). https://doi.org/10.1117/1.600798 . Submission:
JOURNAL ARTICLE
9 PAGES


SHARE
Back to Top