A performance comparison of motion estimation and correction techniques for synthetic aperture sonar imaging

Abstract

Ideally, a synthetic aperture sonar platform traverses a perfectly straight path as it forms the synthetic aperture. However, medium turbulence and currents affect platform motion. Errors between ideal trajectory and true trajectory of a fraction of an acoustic wavelength will cause a loss in phase coherence during data acquisition resulting in unfocused images. Autofocus algorithms, commonly used in synthetic aperture radar (SAR) imaging, have been developed to overcome phase-error problems. However, phase errors in synthetic aperture sonar data are different in nature due to receiver configurations and pulse repetition frequencies. Synthetic aperture radars typically use a single receiver and a high pulse repetition frequency producing a smoothly varying phase error over the synthetic aperture. To achieve satisfactory mapping rates and maximum range, a synthetic aperture sonar receiver needs to use a multiple-element receiver and lower pulse repetition frequency. In synthetic aperture sonar, the resulting phase errors between adjacent receiver elements are typically small and linear, however, phase errors between echo data obtained on successive pulses are significant. An autofocus algorithm has been developed specifically for correcting phase errors in synthetic aperture sonar data. In this paper, its performance is compared with conventional SAR autofocus algorithms on both simulated and experimental data. [This research is supported by UC MICRO program and Sonatech, Inc.]