Optimal feature extraction techniques to improve classification performance, with application to sonar signals

Author

Larkin, Michael J.

Author_Institution

Naval Underwater Warfare Center, Newport, RI, USA

fYear

1997

fDate

24-26 Sep 1997

Firstpage

64

Lastpage

71

Abstract

Feature extraction is an important preliminary step to classification of complex signals. By reducing a high-dimensional signal to a lower-dimensional feature set which preserves the relevant structure of the signal, classification performance is enhanced. A classification system was developed to classify sonar signals as to whether the object detected is minelike or nonminelike. Results are presented comparing classification performance when various feature extraction methods are implemented

Keywords

feature extraction; object detection; optimisation; pattern classification; sonar target recognition; weapons; classification performance; high-dimensional signal; lower-dimensional feature set; minesweeping; object detection; optimal feature extraction techniques; sonar signals; underwater mine detection; Acoustic signal detection; Feature extraction; Government; Neural networks; Object detection; Pattern classification; Signal processing; Sonar applications; Sonar detection; Underwater acoustics;

fLanguage

English

Publisher

ieee

Conference_Titel

Neural Networks for Signal Processing [1997] VII. Proceedings of the 1997 IEEE Workshop

Conference_Location

Amelia Island, FL

ISSN

1089-3555

Print_ISBN

0-7803-4256-9

Type

conf

DOI

10.1109/NNSP.1997.622384

Filename

622384