One of the key challenges for catalogue maintenance of space debris is to correctly correlate new measurements to their originating objects. This paper describes a workflow to associate optical observations to object for which a prior information is available in a robust manner. For efficiency, observations are compressed into atributable vectors and compared against known objects in the measurement space. A pre-filter is used to early eliminate non feasible candidates based on physical constraints and on information theoretic metrics. Suitable candidates are used in a Multiple Hypothesis Bayesian framework to perform the association based on the highest likelihood. The use of a Multiple Hypothesis Filter improves the robustness of the process for the case of ambiguous association of closely spaced objects. Different pre-filter algorithms are described. Results of applying this workflow are shown for GEO and MEO targets.