Sea noise is a non-stationary random series. The acoustic channel noise of shallow sea has the features of remarkable nonlinear and non-Gaussian characteristics of pulse. We use wavelet packets to analyse this kind of noisy signal, and study on different wavelet basis, threshold functions and values of the threshold which will influence the extraction of a LFM signal under the background of strong sea noise. Computer simulation is completed and the best values of the key parameters mentioned above are determined. The experimental results show that if we choose key parameters referred in this paper, we can remove marine acoustic noises effectively and extract the weak LFM signal, by nonlinear filtering based on threshold in wavelet domain.
展开▼