SAS image coherence in the wavelet domain: enhancement of the broadband target response and evaluation of the seafloor complexity

A method is presented for separating the broadband acoustic target response from the background seafloor reverberation based on the multi-look synthetic aperture sonar (SAS) image coherence. The method computes the coherence in the wavelet domain and uses this to distinguish between the target and background, which are assumed to be comparatively coherent and incoherent respectively. A wavelet shrinkage technique is applied to perform the separation whilst preserving phase and image resolution. The method is demonstrated on experimental data from a low-frequency broadband sonar developed for detecting buried objects. We also present preliminary results using a similar approach to obtain a coherence-based metric for seafloor complexity. Results are shown from a high-frequency imaging sonar