Share Email Print

Proceedings Paper • Open Access

Extensions of lp-norm optimum filters for image recognition
Author(s): Nasser Towghi; Bahram Javidi

Paper Abstract

A family of linear and nonlinear processors (filters) for image recognition, which are extensions of the previously developed filters called lp-norm optimum filters, are presented. These filters are lp-norm optimal in terms of tolerance to input noise and discrimination capabilities. The lp-norm is the generalization of the usual mean squared (l2) norm, obtained by replacing the exponent 2 by any positive constant p (usually p ≥ 1). These processors are developed by minimizing the lp-norm of the filter output due to the input scene and the output due to input noise. The minimization is carried out by constraining a function of the filter output to attain a fixed peak value when the input is the target to be detected.

The use of lp-norm to measure the size of the filter output due to noise gives a greater freedom in adjusting the noise robustness and discrimination capabilities. The flexibility in allowing more general type of constraints allows for experimenting and may lead to designing of filters to obtain better performance by selecting an appropriate filter constraint equation to match the metric used to measure the performance of the filter.

we give an unified theoretical basis for developing these filters. This family of filters include some of the existing linear and nonlinear filters.

Paper Details

Date Published: 2 June 1999
PDF: 24 pages
Proc. SPIE 10296, 1999 Euro-American Workshop Optoelectronic Information Processing: A Critical Review, 102960B (2 June 1999); doi: 10.1117/12.365908
Show Author Affiliations
Nasser Towghi, Univ. of Connecticut (United States)
Bahram Javidi, Univ. of Connecticut (United States)

Published in SPIE Proceedings Vol. 10296:
1999 Euro-American Workshop Optoelectronic Information Processing: A Critical Review
Philippe Refregier; Bahram Javidi, Editor(s)

© SPIE. Terms of Use
Back to Top