Reverberation-aware adaptive sparse subspace tracking for moving small target detection in shallow water

To address the problem that moving small-target echoes are easily submerged by strong shallow-water reverberation under fixed or slowly moving platforms, a reverberation-aware adaptive sparse subspace tracking method is proposed. The method exploits the difference between the low-rank coherence of reverberation background and the local sparsity of targets with relative motion. A spatial weighting matrix is constructed from local background power statistics to reduce the influence of strong reverberation regions on subspace estimation. The background subspace is tracked through frame-by-frame recursive updating, and a column-wise sparse constraint is introduced to suppress contamination of the background bases by target highlights and abnormal scatterers. Target responses are then extracted from the orthogonal-complement residual. Simulation results show that the proposed method improves overall detection performance and reduces false alarms caused by reverberation residuals under low signal-to-noise ratio conditions. Lake-trial and shallow-sea trial results further demonstrate that the proposed method effectively suppresses reverberation residuals and improves moving small-target detection performance. The average processing time per frame on the lake-trial data is about 88 ms, supporting frame-by-frame online processing.