Gammatone subband denoising and HHT-based feature extraction for underwater targets

Underwater target recognition, as a key technique, places a great role in underwater acoustic detection. The inevitable noises decrease the performance of the system in the complex underwater acoustic environments. An underwater target recognition algorithm which includes Gammatone filter bank, wavelet threshold denoising and Hilbert-Huang transform (HHT) is proposed based on auditory perception mechanism. Gammatone filter bank is used for the simulation of auditory perception, and the wavelet threshold denoising is applied to enhance the noise robustness of the system. Hilbert-Huang transform is employed as the time frequency analysis tool and used for feature extraction. At last, the efficiency of the proposed algorithm is testified by using the measured underwater target data in the recognition experiment. The results show that the proposed method has a robust performance and good accuracy under the condition of low SNR.