Autofocusing circular synthetic aperture sonar imagery using phase corrections modeled as generalized cones.

Circular synthetic aperture sonar (CSAS) is a coherent aperture synthesis technique that utilizes backscattered acoustic information from an encircled scene to generate information rich, high-resolution imagery. The aperture length required for image synthesis is much longer than in its linear synthetic aperture sonar counterpart and can result in challenging phase delay and navigation estimation constraints. Residual uncorrected phase errors manifest as focus aberrations in reconstructed CSAS imagery. This paper demonstrates that phase error in image patches can be approximated as an aspect variant linear phase shift representable as a generalized cone in wave-number space. If the geometry of the generalized cone is known, it can be applied as the spectral phase of an inverse filter for aberration correction. A method is derived for reconstructing the error cone geometry from independent estimates of its local curvatures, which are found via a series of one-dimensional line searches that maximize the focus of CSAS sub-aperture images. This approach is applied to real and simulated CSAS data containing aperture distortions, and the results successfully demonstrate estimation and correction of the underlying focus aberrations.