Stochastic model-based discrimination of unresolved space objects at the sensor level

Typical algorithms for processing unresolved space imagery from optical systems make restrictive assumptions about the expected behavior of the sensor during collection. Under these assumptions, the geometry of a particular signal usually reveals whether it corresponds to an object of interest. These techniques reduce the number of sensors able to contribute imagery to only those that are directly tasked. The techniques developed in this paper are an effort to derive a general, sensor-level discrimination metric for non-stars based on the expected motion of a star. The result is a probabilistic argument for unresolved signal classification that extends to more general sensor motion profiles than traditional methods.