Adaptive signal detection in compound-Gaussian clutter with inverse Gaussian texture

In this paper, we deal with the problem of signal detection in compound-Gaussian clutter, where the texture is modeled as a random variable with inverse Gaussian distribution. A generalized likelihood ratio test detector for compound-Gaussian clutter with inverse Gaussian texture (GLRT-IG) is presented by a two-step procedure. First the covariance matrix of the speckle is assumed to be known and the statistic test is derived. Subsequently, the estimate of the covariance matrix is substituted into the statistic test. At the performance assessment stage, the influences of both the sample support and the shape parameter on the detector are discussed. Also the comparison with the normalized matched filter shows that the GLRT-IG is a better detector.