Passive Sonar Multi-Target Frequency Feature-Aided Data Association Algorithm Based On DPGMM

In passive sonar system, it's always a challenge to accomplish multi-target data association. We propose a remedy for this problem by using a frequency feature-aided data association algorithm to track targets' frequency and bearing information simultaneously. Frequency and bearing information are extracted by Dirichlet Process Gaussian Mixture Model(DPGMM). Simulation results are presented to illustrate the tracking performance of the proposed algorithm in terms of incorrect association, association interrupt times and tracking accuracy aspect.