Title :
Wavelet transform as a preprocessing method for neural classification of passive sonar signals
Author :
Seixas, J.M. ; Damazio, D.O. ; Diniz, PS R. ; Soares-Fillho, W.
Author_Institution :
COPPE, Univ. Fed. do Rio de Janeiro, Brazil
fDate :
6/23/1905 12:00:00 AM
Abstract :
A neural classifier is developed for passive sonar signals. For achieving data compaction and high performance on the identification of ship classes, the neural processing is performed on preprocessed data in the frequency domain. Preprocessing comprises averaged spectral analysis over contiguous acquisition windows, background noise estimation and wavelet transformation. The overall discrimination efficiency achieved was better than 94%, considering four classes of ships
Keywords :
acoustic noise; frequency-domain analysis; military computing; naval engineering computing; neural nets; object recognition; parameter estimation; signal classification; sonar signal processing; sonar target recognition; spectral analysis; wavelet transforms; averaged spectral analysis; background noise estimation; contiguous acquisition windows; data compaction; discrimination efficiency; frequency domain processing; neural classification; neural classifier; neural processing; passive sonar signals; preprocessed data; ship class identification performance; wavelet transform preprocessing method; wavelet transformation; Acoustic noise; Background noise; Frequency domain analysis; Marine vehicles; Neural networks; Signal processing; Sonar applications; Spectral analysis; Wavelet analysis; Wavelet transforms;
Conference_Titel :
Electronics, Circuits and Systems, 2001. ICECS 2001. The 8th IEEE International Conference on
Print_ISBN :
0-7803-7057-0
DOI :
10.1109/ICECS.2001.957677
