Accession Number:

ADA207210

Title:

High Efficiency Pattern Recognition

Descriptive Note:

Final technical rept. Mar 1985-Feb 1987

Corporate Author:

CALIFORNIA UNIV SAN DIEGO LA JOLLA

Personal Author(s):

Report Date:

1989-01-01

Pagination or Media Count:

10.0

Abstract:

This contract had two main goals the improvement and comparison of optical pattern recognition filter algorithms and the improvement of the optical or Horner efficiency in order to allow smaller lower powered lasers for pattern recognition systems. The first goals was achieved by studying and comparing the performance results of three generic classes of filters the Fukunaga-Koontz, the Hotelling, and the Foley-Sammon filter. All of these are of the classical statistical feature extraction type. In addition the effect of white noise on the recognition of a 2 class problem employing the LSLMT algorithm was evaluated. The second goal was apparently not achieved, as the simulated optical efficiency and classification reliability remained about the same as the full phase and amplitude filters. This may have been due to non- standard definition of Horner efficiency used by the contractor, and the fact that a limited database was used. However, the phase-only filters did show increased peak correlation intensity and SNR.

Subject Categories:

  • Atomic and Molecular Physics and Spectroscopy

Distribution Statement:

APPROVED FOR PUBLIC RELEASE