An Analysis Of Features Of Active Sonar Echoes From Underwater Target

Currently there are some critical problems in extracting features of active sonar echoes from underwater silent small targets,including the stability of the features of target echoes in complex environment,the analysis of time sequence structures of geometrical acoustic scattering and elastic acoustic scattering for the target,the relationship between the physical attribute and the features of target echoes,seafloor reverberation suppression and so on.Therefore,the robust approaches of extracting the features of target echoes for the classic target models in reverberation background,the extracting methods of the independent target acoustic scattering component,and the seafloor reverberation suppression schemes based on the blind separation and the image processing of time-frequency distribution are proposed in this paper,which facilitate the active sonar detection and identification of the underwater silent small targets.Firstly,the approach of extracting the instantaneous frequency features of the target echoes is proposed for improving the stability of the features of target echoes in reverberation background,and the distributions of target echoes and reverberation in feature space are also studied in this paper.The reverberation is considered as a kind of signal with independent characteristics and features.The time-frequency distributions of the target geometry acoustic scattering and reverberation are derived.Based on the randomness of the instantaneous frequency sequence of the signal,four features are extracted,including the variance of instantaneous frequency,the entropy of instantaneous frequency,the kurtosis of instantaneous frequency,and the half-power width of WVD-Radon transformation.Two-dimensional signal feature space is set up by adopting the feature selection method.The results of processing the data from the lake test of detecting the bottom targets show that the distributions of target echoes and reverberation in feature space are stable.Secondly,a signal separation method of target acoustic scattering is proposed for solving the mix between the geometry acoustic scattering and elastic acoustic scattering.Considering a case that LFM signal is transmitted,the target acoustic scattering can be mapped as multiple single frequency components.The linear relationship between the time sequence structures of target acoustic scattering and the mapping components is derived.Thus the specific single frequency component can be obtained by the narrowband filtering.The performance of extracting the independent features of elastic acoustic scattering for the target is verified in the simulations and pool experimental results.Furthermore,the procedure of the delay estimation for the target acoustic scattering is proposed,which suppressing the strong components of the acoustic scattering by band stop filter to improve the weak components of the acoustic scattering,and obtains the frequency by high-resolution spectrum estimation.Pool experimental results show that the delay of the weak edge scattering components in the target echoes can be extracted based on the proposed method.Finally,two aspects of reverberation suppression are studied.The first one is to adopt the morphological filtering to process the time-frequency distribution image of the target echoes for reverberation suppression,considering the problem that the target echoes and reverberation mix with each other in time-frequency distribution space.An approach of structure element design associating with the time-frequency distribution of target geometry acoustic scattering is proposed.The second one is to adopt the variance of instantaneous frequency and the entropy of instantaneous frequency as the target parameters to decide the order of separation results of the target echoes.The performances of the different blind separation methods for separating the target echoes and reverberation are also evaluated.The experimental results from the lake test of detecting the bottom targets and the sea test of detecting the buried targets demonstrate the performance of the proposed methods.