Persymmetric Adaptive Detection for Radar Targets in Compound Gaussian Clutter With Nakagami Texture

The paper focuses on developing adaptive detectors for radar targets in non-homogeneous and non-Gaussian sea clutter. The non-Gaussian characteristics are captured through the compound Gaussian representation with Nakagami-distributed texture components and an unspecified speckle covariance structure. To improve detection reliability when facing limited training samples in non-homogeneous environments, the speckle covariance matrix is modeled with persymmetric constraints, thereby reducing training data dependence. The received radar data are first transformed through the property of the persymmetric matrix and vector. Based on two-step suboptimal test frameworks (including generalized likelihood ratio, Wald, Rao, Durbin, and Gradient tests) and analytical mathematical expressions for estimating clutter parameters, we develop persymmetric adaptive coherent detectors for radar target detection in compound Gaussian clutter with Nakagami-distributed texture. Notably, the test statistics derived from the two-step Gradient, Durbin, Rao tests are shown to be identical. Experimental results employing both simulated and measured radar datasets demonstrate the superior performance of the proposed detectors relative to conventional methods.