A comparison of neural networks and statistical methods for track association in over the horizon radar

An ionospheric model-free pattern classification approach is proposed for associating tracks in over the horizon radar. A set of track features and track affinity measures are derived according to human perceptual grouping principles. To facilitate the pairwise association of the tracks, neural networks and statistical methods are applied to combine different track affinities. A posterior pseudo-probability measuring association is produced for every pair of tracks.