LAUSR

A new multifeature joint detection (MFJD) algorithm is proposed for moving target detection in video synthetic aperture radar (SAR), which uses the spatio-temporal information of a moving target. In spatial dimension, a moving target’s three features, i.e., shadow and unfocused energy in image domain and Doppler energy in the range-Doppler (RD) domain, are predetected using the you only look once (YOLO) algorithm. In the temporal dimension, these three features are jointly tracked by combining the inter-frame information. Subsequently, the corresponding trajectories are matched and correlated. The false alarm and missed alarm in predetection are significantly suppressed by the proposed algorithm in comparison with other existing methods, which have been verified by both the simulation and measured data.