Clutter scenes are nonstationary in space and in time. In order to model clutter environments for extended spatial regions over lengthy time intervals, it is necessary to address the requirements for extended dynamic nonstationary scene generation. This capability is important for testing and evaluating clutter mitigation algorithms that employ spatiotemporal filtering techniques. Traditional Fourier-based and Autoregressive/Moving Average-based (ARMA-based) methods are compared and contrasted for their utility in meeting the objectives of dynamic scene generation. It is shown that Fourier-based approaches can be computationally intensive and that ARMA-based approaches require long warmup periods for structures having long temporal correlations. An alternative approach is presented referred to as the MEMERS (Multiple Extended Moving Evolving Radiance Screens) technique that is computation efficient and satisfies the specified nonstationary spatiotemporal clutter statistics. The MEMERS technique uses multiple radiance screen projection planes to achieve angle decorrelation of clutter scenes. It is shown that by applying motion vectors and imposing time evolution to each projection plane, dynamic clutter scenes having the proper nonstationary spatiotemporal statistics with proper space, time, and angle correlations is achieved.
R. S. Benson,
"Development of ultra-violet plume signature prediction code (PRUV)", Proc. SPIE 1725, Targets, Backgrounds, and Discrimination, 17250B (29 January 1991); doi: 10.1117/12.2300211; https://doi.org/10.1117/12.2300211