Atomic decomposition of geometric acoustic scattering from underwater target

Underwater target detection and recognition is currently a key technology of intelligent acoustic equipment. In the active detection of quiet small underwater targets, an aliasing of distortions between highlight echoes exists in both time and frequency domains. The interference of cross-terms from multi-components in time-frequency domain results in a decline in resolutions of highlight. To solve the problem, the chirplet atom decomposition method is applied to the extraction of geometric highlights when using the linear frequency modulation signals based on the theory of highlight model of echoes from sonar target. The orthogonal matching pursuit algorithm is also added to obtain the time-frequency form of the highlight echoes without cross-term interference. The analytical result of the simulation and data processing are consistent with the theoretical acoustic scattering highlight model, which proves that the proposed method can effectively extract the characteristics of underwater target geometric highlights and improve the detection and identification ability of underwater targets.