Geometry-Aware Entropy Minimization Autofocusing Algorithm for UAV-Based SAR

Synthetic Aperture Radar (SAR) systems deployed on Unmanned Aerial Vehicles (UAVs) are emerging as a highly promising solution for a wide range of future remote sensing and monitoring applications. The integration of SAR’s well-established capabilities for high-resolution imaging and change detection with the flexibility, maneuverability, and cost-effectiveness of modern UAV platforms unlocks significant potential for innovative sensing missions. However, the practical implementation of SAR on UAVs presents considerable challenges, primarily due to the inherent limitations in UAV flight stability and precise navigation, which degrade image quality. As a result, robust autofocusing techniques are essential to correct motion-induced phase errors and ensure coherent image reconstruction. While numerous autofocusing methods have been proposed in the literature, most are designed under assumptions that do not hold for low-flying UAV platforms. Specifically, conventional algorithms often overlook the strong dependence of the SAR imaging geometry on both azimuth and range positions of the targets, which becomes pronounced at low altitudes. In this study, we introduce a novel autofocusing approach termed Geometry-Aware Entropy Minimization (GAME). The proposed algorithm reconstructs three-dimensional positioning errors by explicitly modeling the spatially variant acquisition geometry along the UAV trajectory. This geometry-aware formulation enables accurate autofocus correction under the challenging conditions associated with low-altitude UAV SAR acquisitions. The effectiveness of the GAME algorithm is demonstrated using real X-band SAR data collected by an octacopter platform, confirming its applicability and performance in practical scenarios.