A score-based track initiation procedure to eliminate false tracks in the presence of sea clutter

Track initiation methods attempt to start tracks for real targets by eliminating false tracks caused by clutter during target tracking. This process becomes even more critical for sea surveillance in the presence of sea clutter, e.g. waves and ship wakes, since the measurements from clutter are highly likely appear in gates of real targets at track initiation stage. This paper discusses the usage of number of measurements in a score based track initiation algorithm. The measurements are first grouped according to their proximity and then the number of measurements in each group is modeled as K-distributed for clutter and Poisson distributed for true targets. Initiation uses a likelihood ratio which is based on kinematics as well as number of measurements in a group. Real data acquired with experimental radar is used for modeling. This paper discusses the implementation of mentioned Score-based algorithm above and compares its performance against two other methods: M-out-of-N and kinematics-only score-based initiation on real data. Proposed technique produces significantly less number of false tracks compared to others.