Impact of Measurement-to-Track Data Association Errors on RCS-Based Target Classification

摘要

The vast majority of literature on radar cross section (RCS)-based target classification assumes that pure tracks (i.e., tracks in which measurements all originate from the same object) are available. However, this assumption is frequently violated, leading to degraded classification performance. Measurement-to-track data association errors that lead to impure tracks can occur for a number of reasons, including merged measurements (due to unresolved targets, possibly because they are separating or crossing) and track accuracy errors. Certain technology advances, such as wideband waveforms and multiple hypothesis tracking, are expected to mitigate this problem. Even if data association errors become less frequent as such technology is deployed, impure tracks will remain an unpopular fact of life that ought to be addressed.