A Track Registration Association Technique – Feature Analysis

Abstract

A recent track-to-track association technique was developed to associate tracks from multiple sensor platforms where position and heading errors were incorporated into the reports. When this issue, known as target registration, is present, standard association routines often result in misassociations or non-associations between the tracks of the same target. This new technique develops a track representation that is invariant in translation errors and insensitive to rotation errors. It has been demonstrated to be effective for registration problems with reasonable heading errors and any translation errors. In this work, the features used in the track representation are analyzed for their effectiveness and sensitivity for various registration error sizes, providing the foundation for the weighting of the features to enhance registration association.