Share Email Print

Proceedings Paper

Application of canonical correlation analysis in detection in presence of spatially correlated noise
Author(s): Wei Guo Chen; James P. Reilly; Kon Max Wong
Format Member Price Non-Member Price
PDF $14.40 $18.00

Paper Abstract

A new approach to the problem of detecting the number of signals in unknown colored noise environments is presented. Based on an assumption that the noise is correlated only over a limited spatial range, the principle of canonical correlation analysis is applied to the outputs of two spatially-separated arrays. The number of signals is determined by testing the significance of the sample canonical correlation coefficients. The new method is shown to work well in both white and unknown colored noise situations and does not require any subjective threshold setting. Instead, a set of threshold values are generated according to a specified or desired false alarm rate. Simulation results are included to illustrate the comparative performance of the proposed canonical correlation technique (CCT), versus the well-known AIC and MDL criteria, in colored noise. It is found that the performance of the AIC and MDL criteria degrade very rapidly as the degree of color in the noise increases. On the other hand, the performance of the CCT method is relatively insensitive with respect to variations in degree of color.

Paper Details

Date Published: 1 December 1991
PDF: 12 pages
Proc. SPIE 1566, Advanced Signal Processing Algorithms, Architectures, and Implementations II, (1 December 1991); doi: 10.1117/12.49845
Show Author Affiliations
Wei Guo Chen, McMaster Univ. (Canada)
James P. Reilly, McMaster Univ. (Canada)
Kon Max Wong, McMaster Univ. (Canada)

Published in SPIE Proceedings Vol. 1566:
Advanced Signal Processing Algorithms, Architectures, and Implementations II
Franklin T. Luk, Editor(s)

© SPIE. Terms of Use
Back to Top