A Doppler-compensated framework for SAR–AIS ship association and ship–wake-based speed inversion using medium-resolution SAR

SAR–AIS data fusion is essential for maritime surveillance, but Doppler-induced displacement of moving ships causes systematic inconsistencies between AIS-interpolated positions and SAR-detected ship signatures. This study proposes a Doppler-compensated framework for SAR–AIS ship association and ship–wake-based speed inversion using medium-resolution Sentinel-1 SAR imagery. The method corrects the dominant Doppler displacement using AIS motion and SAR imaging geometry, refines residual spatial bias by CPD point-set registration, and obtains final one-to-one associations through multi-feature Hungarian assignment. Ship speed is then inverted from ship–wake displacement, and a confidence index Q v is developed to evaluate sample-level inversion reliability. Experiments using Sentinel-1 imagery, POEORB data, and AIS records from 2017 to 2025 over the Gulf of Mexico and the Florida Straits show that the median SAR–AIS association distance decreases from 414.84 m to 91.37 m. For 2514 valid samples, speed inversion achieves an MAE of 4.503 kn and an RMSE of 6.687 kn; for high-confidence samples with Q v > 0.7 , the MAE decreases to 2.331 kn and the RMSE to 3.208 kn.