A pattern recognition Wiener filter for realistic clutter backgrounds

• Published in: Optics communications Vol. 172; no. 1-6; pp. 193 - 202
• Main Authors: TAN, S, YOUNG, R. C. D, BUDGETT, D. M, RICHARDSON, J. D, CHATWIN, C. R
• Format: Journal Article
• Language: English
• Published: Amsterdam Elsevier Science 15.12.1999
• Subjects: Computers in experimental physics; Exact sciences and technology; Fundamental areas of phenomenology (including applications); Image processing; Imaging and optical processing; Instruments, apparatus, components and techniques common to several branches of physics and astronomy; Optical elements, devices, and systems; Optical processors, correlators, and modulators; Optics; Pattern recognition; Physics; Theoretical study; Matched filters; Optical correlator; Wiener filters; Pattern recognition; Noise
• ISSN: 0030-4018; 1873-0310
• DOI: 10.1016/S0030-4018(99)00599-4

The Wiener filter can be used in place of the matched filter in correlation based pattern recognition. It has been shown to be capable of detecting a target in severe clutter backgrounds provided the clutter power spectrum is included in the filter transfer function. It is shown here that the performance degrades very markedly if a different background from that used to create the filter is present at the input. However, multiple input scene power spectrum averaging is shown to considerably improve the performance, even when previously unknown backgrounds are encountered that have not been included in the averaged spectrum. This leads to the generation of a clutter power spectrum model and it is shown that use of this in the filter transfer function allows much improved clutter resistance when unknown background scenes are encountered, improving the prospects for the practical application of the filter to demanding pattern recognition problems.