High-resolution sonar imaging based on wideband fast iterative shrinkage thresholding deconvolution beamforming

Conventional beamforming (CBF) has been widely applied for underwater acoustic imaging owing to its robustness and low complexity. However, its resolution is limited by the Rayleigh threshold. Thus, various imaging deblurring methods, such as deconvolution beamforming, have been proposed to improve resolution without changing the array aperture. But these intensity-based methods require the targets to be incoherent and are based on the narrowband signals, which mismatches with practical sonar imaging model. In response, this study presents a matrix transformation and monotone fast iterative shrinkage thresholding algorithm (MFISTA) to extend deconvolution beamforming to wideband coherent signals for high-resolution imaging. Wherein the former focuses received signals to a reference frequency by the focusing signal subspace, which can reasonably approximate a broadband signal as a narrowband. Then the MFISTA develops a complex-domain deconvolution beamforming to theoretically remove the cross term interference generated by coherent targets in intensity-based methods. Simulations in uniform linear array demonstrate that the main lobewidth of the proposed method is less than 1/4 that of the CBF method and approximately 1/3 that of other CBFs, while its sidelobe level is over 5 dB lower than other methods. Lake-based imaging experiments validated that the proposed method has superior resolution, main-lobewidth, sidelobe level, and noise immunity compared to other intensity-based methods.