Simulation of a Multitarget, Multisensor, Track-Splitting Tracker For Maritime Surveillance

摘要

This study adapts some established target tracking techniques for use in the maritime surface surveillance role and tests them with computer generated data. Computer simulation of a track splitting tracker capable of operating in this undersampled and asynchronous environment is presented. The tracker uses standard and extended Kalman Filter algorithms to estimate target state from latitude and longitude or line of bearing position measurements. Prior to state estimation, all measurements are processed to retain only those that meet feature and geographic gate thresholds. All measurements passing these criteria will update the target state and be scored based on a goodness of fit with the model. The state estimate with the best score is selected as the correct one for display purposes, while all state estimates continue to be processed with subsequent measurements. Several runs of the simulation are discussed here to illustrate the performance of track splitting and the effect of several key tracker parameters.