Accession Number:

ADA149562

Title:

Optical Implementation of the Synthetic Discriminant Function,

Descriptive Note:

Corporate Author:

AIR FORCE ARMAMENT LAB EGLIN AFB FL

Personal Author(s):

Report Date:

1984-10-21

Pagination or Media Count:

9.0

Abstract:

Much attention is focused on the use of coherent optical pattern recognition OPR using matched spatial filters for robotics and intelligent systems. The OPR problem consists of three aspects -- information input, information processing, and information output. This paper discusses the information processing aspect which consists of choosing a filter to provide robust correlation with high efficiency. The filter should ideally be invariant to image shift, rotation and scale, provide a reasonable signal-to-noise SN ratio and allow high throughput efficiency. The physical implementation of a spatial matched filter involves many choices. These include the use of conventional holograms or computer-generated holograms CGH and utilizing absorption or phase materials. Conventional holograms inherently modify the reference image by non-uniform emphasis of spatial frequencies. Proper use of film nonlinearity provides improved filter performance by emphasizing frequency ranges crucial to target discrimination. In the case of a CGH, the emphasis of the reference magnitude and phase can be controlled independently of the continuous tone or binary writing processes. This paper describes computer simulation and optical implementation of a geometrical shape and a Synthetic Discriminant Function SDF matched filter. The authors chose the binary Allebach-Keegan AK CGH algorithm to produce actual filters. The performances of these filters were measured to verify the simulation results. This paper provides a brief summary of the matched filter theory, the SDF, CGH algorithms, Phase-Only-Filtering, simulation procedures, and results.

Subject Categories:

  • Theoretical Mathematics
  • Optics

Distribution Statement:

APPROVED FOR PUBLIC RELEASE