Estimation of non-Rayleigh clutter and fluctuating-target models

The detection performance prediction for radar and sonar systems requires statistical models of the matched filter output for the ´clutter-only´ and ´target-plus-clutter´ cases. Two practical techniques are presented for deriving these models from matched filter output data. The first technique is based upon a generalised Gram-Charlier series expansion, whose terms consist of the generalised gamma probability density function and its derivatives. The second technique is based on the Rayleigh mixture model and it is shown that the target-plus-clutter case can be expressed as a convolution of mixture parameters of the separate target-only and clutter-only cases. Numerical examples are presented.