A hybrid time and wavenumber domain algorithm for generalized beamforming with circular synthetic aperture data

Abstract

Synthetic aperture sonar systems typically scan along a linear trajectory, however autonomous underwater vehicles (AUV's) provide a useful platform for performing scans along more complicated trajectories with specific advantages. One example is the circular scan, which provides superior wavenumber coverage and resolution for images reconstructed near the circle center. A variety of approaches have been put forward for beamforming synthetic aperture data collected along a circular trajectory, but for images that span a significant fraction of the circle radius, accelerated time-domain approaches such as fast-factorized backprojection (FFBP) have been suggested. The computational requirements for generating a fully sampled image collected around a circular aperture are extremely large, however, even for FFBP. This is primarily because of the extremely small pixel size required to maintain adequate spatial sampling as larger portions of the aperture are coherently fused in the later stages of the FFBP algorithm. A method is presented for replacing the later stages of the algorithm with a series of spectral mappings, retaining the generality of the beamformer while improving computational efficiency.