Correlated K distributions can be used to describe the clutter seen in coherent radar images. To test the performance of detection and classification algorithms in controlled clutter environments it is necessary to be able to produce accurate simulations of the clutter. For this reason, the simulation of correlated K distributed processes is of interest. A new method for simulating such processes is described which involves applying a variety of moving average filters to sequences of independent gamma distributed variables and summing the resulting sequences to obtain the desired autocorrelation properties. Using this technique, it is possible to match the desired single point statistics and correlation properties exactly although the procedure becomes increasingly complex as the correlation length increases. In common with all other currently available techniques for simulating correlated K distributions, the proposed scheme does not solve the problem of introducing anti-correlations which remains a subject of research. Simulations of both one- and two-dimensional processes are presented which demonstrate the accuracy of the technique.
展开▼
