Classification of clutter using the bispectrum

Three statistically independent time series associated with three forms of clutter, namely ground, weather and bird, and sea clutter are being classified based on their bicoherences. The key feature used for identifying clutter is the skewness of each clutter distribution. The performance of the proposed bispectral based classifier is compared with that of spectral based classifiers. Scenarios entailing different combinations of clutter are also being examined under the assumption that all clutter scattering features are corrupted with additive white Gaussian noise.<>