Abstract: In this paper, a new method of phase errors estimation based on higher-order statistics is proposed forSAR imagery autofocus. The method, which starts right in with complex phase-degraded SAR imagery, isolates thedominant point target in the image domain via circular shifting and windowing, and then estimates phase errors inthe range-compressed phase-history domain by a higher-order cumulant . The algorithm makes good use of the property that the higher-order cumulant can suppress noise with Gaussian distribution, while avoiding derivative computation which can be sensitive to additive noise and disturbance. So the phase errors estimation will be more robust .Simulation and processing results of real data show the feasibility of the proposed method.

Key words: Synthetic Aperture Radar (SAR), autofocus, higher order statistics, phase errors, Phase Gradient Autofocus (PGA)