OPTIMUM CLASSIFICATION OF NON-GAUSSIAN PROCESSES USING NEURAL NETWORKS

A prerequisite for target detection in synthetic aperture radar and mo ving target imaging radars is an ability to classify background clutte r in an optimal manner. Such radar clutter can frequently be modelled as a correlated non-Gaussian process with, for example, Weibull or K s tatistics. Maximum likelihood (ML) provides an optimum classification scheme but cannot always be formulated when correlations are present. In such circumstances, nonlinear, adaptive filters are required which can learn to classify the clutter types: a role to which neural networ ks are particularly suited. The authors investigate how closely neural networks can approach optimum classification. To this end, a factoris ation technique is presented which aids convergence to the best possib le solution obtainable from the training data. The performances of fac torised networks are compared with the ML performance and the performa nces of various intuitive and approximate classification schemes when applied to uncorrelated K distributed images. Furthermore, preliminary results are presented for the classification of correlated processes. It is seen that factorised neural networks can produce an accurate nu merical approximation to the ML solution and will thus be of great ben efit in radar clutter classification.