Target spectra feature selection using a priori information
Abstract
The current work presents a method of selecting the minimum number of features out of a given number of fast Fourier transform filter outputs capable of characterizing a given number of target signatures to be recognized. The method uses a priori information on the target dimensions. It consists of minimizing an upper bound on the error probability, which is a sum of the Bhattacharyya coefficients weighted by the a priori probabilities. The reason for employing this approach is that the 'fuzzy' a priori information can then be used by modifying the a priori probabilities in the equations for the upper bound.
- Publication:
-
IEEE Transactions on Aerospace Electronic Systems
- Pub Date:
- July 1975
- DOI:
- 10.1109/TAES.1975.308133
- Bibcode:
- 1975ITAES..11..650R
- Keywords:
-
- Bayes Theorem;
- Fast Fourier Transformations;
- Radar Signatures;
- Signature Analysis;
- Target Recognition;
- Classifying;
- Radar Targets;
- Sonar;
- Spectrum Analysis;
- Communications and Radar