Abstract: The key of underwater bottom target recognition is to extract the stable target echo signal feature from reverberation background. The edge scattering of target carries its geometrical information, and it has regular distribution on the time-frequency plane. But the path difference of edge scattering is little, and it’s always drowned out by reverberation. Because of the limited time-frequency resolution and the limited anti-reverberation ability of existing time-frequency analysis method, there hasn’t been any effective feature extraction method of target geometrical characteristics. The geometrical distribution feature of target echo on the time-frequency plane is extracted as an image feature. The morphological characteristics of geometrical highlight and reverberation on the time-frequency plane are researched. The structural element matched with the time-frequency distribution is constructed, and the time-frequency distribution image is morphologically filtered. The experimental data processing result shows that, compared with existing method, the proposed method can be used to recognize the geometrical highlight structure of target and suppress the reverberation at the same time.

Key words: acoustics, bottom target, geometrical highlight feature, anti-reverberation, morphological filter