Optimized Algorithms for Detection of Sparse Targets in Heterogeneous Gaussian Noise

In this paper we propose two adaptive detection algorithms for sparse targets embedded in heterogeneous AR Gaussian noise. The first one solves the problem of estimating the subset of cells containing a scatterer via the GLRT principle, while the latter models the number of scatterers as a random parameter and relies on the use of quantized statistics. A preliminary performance assessment, conducted by Monte Carlo simulation, has shown that both solutions allow to reduce the detrimental effects, in terms of collapsing loss, suffered by conventional solutions. In particular the former algorithm is to be preferred in terms of performance while the latter has a lower computational complexity.