A Novel Merging Algorithm in Gaussian Mixture Probability Hypothesis Density Filter for Close Proximity Targets Tracking

摘要

This paper proposes a novel merging algorithm in Gaussian mixture probability hypothesis density filter to track close proximity targets. The proposed algorithm is added after GM-PHD recursion, in a condition that more than one target has the same state. The weights of Gaussian components decide whether the components can be utilized to extract states, and the means and covariances of Gaussian components are used to determine the distance of components. Depending on these weights, means and covariances, the proposed algorithm avoids that the components which have higher weights than other components are merged in foresaid condition. Simulation results show that the new algorithm can enhance the precision of estimation for multi-target states when the targets move closely.