Research On VTS Target Tracking Approach Based On Random Finite Set Theory

Target tracking is to estimate the target number and the associated states in cluttered environments, which has been widely applied in transportation, radar information processing, video surveillance, etc. The traditional target tracking approaches conclude the data association as well as plenty of the prior conditions that are the major constraints on the efficiency and reliability. This dissertation begins with the framework of the Random Finite Set (RFS) theory. Aiming at the nonlinear and non-Gaussian system, a series of the improved filtering algorithms are proposed to track the Vessel Traffic Services (VTS) target. The main work of the dissertation is as follows:(1) For the limitations of α-β filtering algorithm, an improved algorithm is proposed to track the constant velocity target and the constant accelerate target. Then, an improved algorithm is proposed to solve the computational complexity of the Markov transition probability matrix as well as the process noise affected by target maneuvers. The efficiency and reliability of the proposed approaches are illustrated from simulation results.(2) For the undefined detection conditions, the Sequential Monte Carlo (SMC) implementations of the Probability Hypothesis Density (PHD) filters for the unknown detection probability and the unknown clutter are derived respectively. Then, to correct the overestimated target number using the standard PHD filter, an improved algorithm is proposed, which balances the relationship between the number of false alarms and that of true targets. Finally, a novel multi-model PHD filtering algorithm is proposed to enhance the tracking performance as targets maneuver. Meanwhile, the true states are reflected. The efficiency and reliability of the proposed approaches are illustrated from simulation results.(3) For the undefined detection conditions, the SMC implementations of the Cardinalized Probability Hypothesis Density (CPHD) filter for the unknown detection probability and the unknown clutter are derived respectively. Then, to overcome the missed detection using the standard CPHD filter, an improved algorithm is proposed, which reduces the false and missing alarm rate. Finally, a novel multi-model CPHD filtering algorithm is proposed to cut down the number of both the state equation dimensions and the required sampling particles. The efficiency and reliability of the proposed approaches are illustrated from simulation results.(4) For the measurement data received by the VTS radar, the target tracking approaches based on the RFS theory are applied for the VTS radar information processing system to achieve the VTS target number and the associated states on the designed system architecture. Finally, the VTS target tracking in the complicate environments is achieved. The efficiency and reliability of the proposed approaches are illustrated from experiment results.