Echo components for aspect-independent detection and classification

Data from a numerical target scattering model are described in the time and frequency domains. These echoes are separated into physics-based components called wavepackets, which are aspect-dependent. These individual components are made stochastic by the assumption of random aspect. Each physical component is then decomposed with a statistically independent basis via the Karhunen-Loeve Expansion (KLE). It is shown that the echo is well described by a collection of a few terms of the KLE. The statistically independent expansion amplitudes are then used to generate detection statistics.