RADAR TARGET DETECTION IN STRONG CLUTTER USING SPATIAL-TEMPORAL U-NET

In radar target detection problems, the detection performance relies heavily on the signal-to-clutter-ratio (SCR). In this paper, we investigate the problem of radar target detection in strong clutter. When the clutter is extremely strong, it is almost impossible to detect the targets with only one range-Doppler-matrix (RDM). To address this issue, we resort to another time dimension to improve the detection performance, and the Spatial-Temporal U-Net (ST U-Net) is employed. Compared with ordinary deep learning based detectors which only take the RDM under test as the input, the ST U-Net simultaneously takes in the RDM of the current coherent processing interval (CPI) and the neighboring CPIs to jointly learn the spatial and temporal patterns, which leads to better detection performance. Extensive simulation results show that the proposed ST U-Net outperforms other detectors in strong clutter environment.