Angles-only data association using directional discriminant analysis

Suppose that for a library of space objects, their predicted angular positions at the current time are available from previous observations, including an assessment of the errors. Given a new angles-only observation at the current time, the objective is to decide which object, if any, in the libary corresponds to the observed object. The Fisher-Bingham-Kent distribution from directional statistics provides a powerful and tractable model to summarize uncertainty and to implement the classic methods of discriminant analysis from multivariate analysis for this data association problem