Space Object Data Association Using Spatial Pattern Recognition Approaches

Abstract

Identification of known space objects is a critical step in maintaining accurate catalogs for space situational awareness activities. With increasing numbers of objects in orbit, optical measurements of space objects become more populated with detections, stressing the algorithms used to track and identify these objects. Traditional algorithms used for identifying space objects, such as elliptical gating, suffer from ambiguous or incorrect classifications as gates tend to overlap in dense detection environments. An algorithm is developed that couples elliptical gating with a star pattern recognition algorithm called the planar triangle method to overcome the difficulties found in spatially dense observations. Unlike star catalogs, cataloged resident space objects often contain considerable uncertainty, further challenging the identification of objects in a cluttered field-of-view. The proposed approach leverages uncertainties of the catalog as well as the optical measurement sensor uncertainty to support the space object identification. Simulation results using the gating-assisted planar triangle method show a significant improvement in robust identification of space objects as compared to traditional elliptical gating methods when faced with highly cluttered observations.