Compressive sensing for synthetic aperture sonar

Abstract

This work analyzes approaches to synthetic aperture sonar imaging based on the concept of compressive sensing and compressive sampling. In compressed sensing, a low-dimensional linear projection is used to acquire an efficient representation of a compressible signal directly using just a small number of measurements. The signal (and image) is then reconstructed by solving an inverse problem by linear programming. It will be demonstrated that compressed sensing allows for improvements in synthetic aperture sonar systems by potentially reducing array sizes, beamforming computational requirements, synthetic array sampling requirements, image storage limitations, and allow for increasing vehicle speed. Results show that low error rates are feasible with several compressive sensing algorithms.