A hybrid SAR autofocus technique by two methods of sub-aperture estimation and iterative golden section search

Abstract

In a real airborne synthetic aperture radar (SAR), its major phase errors are usually composed of two categories, such as slow-time varying phase errors (less than several cycles of change in phase during synthetic aperture time) and fast-time varying phase errors (otherwise, including wide band random) according to the motion of aircraft. If the fast errors are no more negligible compared to the slow errors, they should be estimated and then compensated accurately to obtain a well focused image. However, it is not proper to estimate all phase errors at the same time like conventional autofocus techniques because the estimation of the fast-time varying phase errors are seriously affected by blurring in image due to the slow-time varying phase errors. In this paper, we presents an accurate hybrid phase estimation technique using two independent estimation stages of sub-aperture and an iterative golden section search method, which has advantages over several existing methods, because of its better estimation accuracy and less sensitive to the quality of extracted range bins as well as requiring less computation time. The performance of our method is illustrated by simulations of point targets and an experiment with real SAR data.