Modelling of inhomogeneity in radar clutter using Weibull mixture densities

In this study, it is demonstrated that inhomogeneity observed in a variety of radar experimental data can be modelled as a finite mixture of Weibull densities. An algorithm for identifying the parameters of each Weibull component in the mixture is developed. The proposed mixture identification algorithm is then applied to three representative sets of radar experimental data, and the quality of the statistical fit is assessed using the p-value, which is estimated based on the Kolmogorov-Smirnoff distance. An application of the Weibull mixture identification algorithm to detect an extended ground target is demonstrated. It is shown that based on the proposed Weibull mixture model, an optimal constant false-alarm rate detector can be implemented even in the presence of inhomogeneity.