Bayesian Space-Time Adaptive Target Detector in Reverberation Edge Environments

Active sonar faces the problem of non-uniform data in reverberation edge environments, leading to a sharp decrease in the amount of auxiliary data consistent with the background statistical characteristics of the cell under test. Meanwhile, existing methods utilize the sample covariance matrix of data when estimating the reverberant covariance matrix, which has limited the estimation accuracy, inevitably resulting in a loss of target detection performance. To address the above issues, this paper conducts research on two-step generalized likelihood ratio detection methods in reverberant edge environments based on Bayesian target detection, proposing two adaptive target detection methods: Bayesian Hermitian Target Detector in Reverberation Edge Environment, BH-TD-REE, and Bayesian Symmetric Target Detector in Reverberation Edge Environment, BS-TD-REE. Both the BH-TD-REE and BS-TD-REE methods treat the reverberation covariance matrix as a random variable and adopt the maximum a posteriori estimation criterion instead of the sample covariance estimation. Furthermore, the BS-TD-REE method also utilizes the spectral symmetry property of the reverberation covariance matrix, effectively expanding the number of auxiliary data and reducing the detector's demand for uniform auxiliary data. Simulation results show that both proposed methods effectively improve target detection performance and robustness, while the detection performance advantage is more significant under limited auxiliary data quantity.