Target Feature Extraction for Passive Sonar Based on Two Cepstrums

摘要

A passive sonar target can be regarded as a sounder from the viewpoint of sonar operators. The impulse response of the sounder in the cepstrum domain, which shows physical features of targets, can be obtained from the radiated-noise of targets using LPC cepstrum and Mel cepstrum. A set of cepstrum-domain features is extracted based on the above two cepstrums of the impulse response. The neural network target classifier was designed using cepstrum-domain features. The classification experiments were carried out for three different kinds of targets based on practical data. The experimental results show that the feature extraction method based on two cepstrums are useful.