A SAR Autofocus Technique With MUSIC and Golden Section Search for Range Bins With Multiple Point Scatterers

Abstract

Many synthetic aperture radar (SAR) autofocus techniques use range bins containing a single dominant point scatterer to estimate the phase error by maximizing or minimizing an objective function. We analytically show that some widely used objective functions do not give accurate phase error estimates if the objective function is constructed using a range bin containing multiple strong point scatterers (SPSs). Multiple SPSs are often observed in the range bins extracted from SAR images of urban areas containing many bright man-made objects. Such multiple SPSs do not allow us to obtain accurate estimates due to the interference between SPSs. To overcome this multiple scatterer problem, we propose the use of a combined entropy objective function with the local magnitudes of SPSs, along with the multiple-signal classification algorithm. Our experiment with actual SAR data confirms the superiority of the proposed approach.