An Improved Backprojection Autofocus Method for Synthetic Aperture Sonar

ABSTRACT Autofocus methods effectively compensate for multi‐source phase errors in synthetic aperture radar (SAR) and synthetic aperture sonar (SAS). Backprojection (BP)‐based autofocus has attracted wide attention for its high flexibility and low implementation constraints. Although sharpness or entropy‐based BP autofocus is widely used in SAR, its application in SAS is rarely reported. Moreover, sharpness‐maximising BP autofocus suffers from a heavy computational burden, as it requires iterative optimal phase error search and extensive pixel‐wise operations. To solve this problem, we propose an improved BP autofocus method with two core acceleration strategies to reduce computational load. First, Anderson acceleration (AA), designed to speed up iterative searches, is applied to autofocus iterations to cut time consumption by reducing required iteration counts. Second, we verify that high‐intensity pixels in SAS images exhibit spatial sparsity and dominate image statistical characteristics. Using only these extracted pixels in iterations greatly reduces computational burden. Experimental results show that, compared with state‐of‐the‐art BP autofocus, the proposed method achieves equivalent image quality while reducing computation time by one to two orders of magnitude.