Application of wavelet transform for classification of underwater acoustic signals

Inspired by the experience of training human experts in sonar, automatic classification of signals detected by sonar is used to recognize the platforms. Many techniques of feature extraction have been developed, such as Mel-frequency cepstrum coefficients, to simulate passive target signal. The paper proposes a method that integrates wavelet transform to the feature extraction method, and the resulting features are employed for the classification problem. The classifier identification rate is calculated, and the performance of the recognition model is evaluated for different range, speed, and direction for the maritime target. Moreover, the performance of the classifier in noisy conditions is investigated.Inspired by the experience of training human experts in sonar, automatic classification of signals detected by sonar is used to recognize the platforms. Many techniques of feature extraction have been developed, such as Mel-frequency cepstrum coefficients, to simulate passive target signal. The paper proposes a method that integrates wavelet transform to the feature extraction method, and the resulting features are employed for the classification problem. The classifier identification rate is calculated, and the performance of the recognition model is evaluated for different range, speed, and direction for the maritime target. Moreover, the performance of the classifier in noisy conditions is investigated.