Expand this Topic clickable element to expand a topic
Skip to content
Optica Publishing Group

Synthesis of distortion invariant optical correlation filters

Open Access Open Access

Abstract

A method is presented for synthesizing optical correlation filters that will detect a target image independent of position and other general kinds of image distortion. The various distortions can be rotation, scale, and perspective, or any combination of these. Most generally, any linear coordinate distortion is allowed. Since these filters do not require energy normalization, they are also intensity invariant. The design technique starts by finding a set of eigenfunctions and eigenimages for the specified target and distortion. Because these modes completely describe the target image, the correlation filter is formed from their linear combination. The spatial orthogonality of the eigenimages preserves the target information when the modes are combined. The linear combination coefficients are required to have unit amplitude, and their phases are found so that the optical system outputs a constant amplitude. Filters meeting these criteria are successfully designed by iterating between the spatial domain and the eigenfunction domain. All the target information can be extracted by the optical system if the distortion is introduced as a temporal variation of the filter. Detection of a constant intensity signature will uniquely locate input targets in parallel. We illustrate this technique with several examples.

© 1986 Optical Society of America

PDF Article
More Like This
Iterative Synthesis of Distortion Invariant Optical Correlation Filters

George F. Schils and Donald W. Sweeney
WB3 Signal Recovery and Synthesis (SRS) 1986

Iterative technique for the synthesis of optical correlation filters

George F. Schils and Donald W. Sweeney
THE6 OSA Annual Meeting (FIO) 1985

Correlation Filters for Distortion-Invariance and Discrimination

David Casasent and Abhijit Mahalanobis
ThD4 Machine Vision (MV) 1985

Select as filters


Select Topics Cancel
© Copyright 2024 | Optica Publishing Group. All rights reserved, including rights for text and data mining and training of artificial technologies or similar technologies.