Time–frequency filtering-based autofocus

Autofocus is very important to the imaging of air-borne synthetic aperture radar (SAR). There have been many kinds of the autofocus technique, which could fall into two categories according to whether the phase error is approximated with a parametric model or not. On the condition that prominent points exist in the image, the non-parametric methods have superiority over the parametric ones in dealing with high-order phase errors due to their high precision and appropriate computational complexity. As will be shown in this paper, the signal-to-clutter ratio (SCR) plays a pivotal roll in this kind of autofocus. The time–frequency filtering is introduced because of its extraordinary performance in raising the SCR of time-variant signals. Since the main part of the phase error is quadratic, the short-time fractional Fourier domain (STFRFD) filtering is utilized, and then a novel autofocus method termed time–frequency filtering-based autofocus (TFFA) is proposed. Theoretical analysis and simulations show the validity of this approach.