A novel Bayesian adaptive detector against non-Gaussian clutter

We present a novel two-step radar signal detector against the non-Gaussian distributed clutter in this paper. The compound Gaussian distribution is employed to model the clutter, which describes the clutter vector as the product of a non-negative random variable (the so-called texture) with a complex Gaussian random vector (called the speckle). The posteriori probability density function of the texture parameter is given by the Bayesian approach using a non-informative priori. Then the expression of testing statistic is derived according to likelihood ratio test rules. The Monte Carlo simulation is taken to investigate the performance of the detector in several kinds of clutter environments and the results show its improvement compared to the conventional adaptive matched filter detector.