A Forward-Squinted Synthetic Aperture Radar Autofocus Algorithm Under Large Trajectory Deviations

This study proposes an improved autofocusing algorithm to address the forward-squinted synthetic aperture radar （SAR） imaging degradation caused by large trajectory deviations. The degradation causes inaccuracies in Doppler ambiguity estimation and generates diverse defocusing patterns， which severely degrade image focus quality. First， a Doppler center estimation method based on range frequency domain sub-band signals is proposed. Baseband Doppler centroids are estimated for sub-band signals via the correlation function， enabling the construction of a binary linear equation system that relates sub-band centroids to Doppler ambiguity numbers determined using least squares （LS） estimation. This approach effectively resolves the inaccuracy in Doppler ambiguity estimation. Second， an adaptive windowing-weighted phase gradient autofocus （WPGA） approach based on superpixel segmentation is presented， accurately segmenting and extracting defocused target regions by employing pixel dissimilarity. Adaptively adjusting the window length according to the dimensions of the superpixel units within the target regions significantly enhances the accuracy of phase error estimation. Simulations and real-data SAR processing results demonstrate that the proposed algorithm maintains superior focusing capability and robustness under large trajectory deviations.