An Novel Multiple Hypothesis Tracker with Adaptive Clutter Estimation for Heterogeneous Scene

In most existing multiple hypothesis trackers (MHT), the clutter intensity is usually assumed to be prior known, and the clutter intensity can be expressed as the product of the probability density of clutter spatial distribution and the average number of false alarms caused by clutter per scan. To improve the target tracking performance of the MHT in heterogeneous scenes, we proposed a new novel MHT algorithm with clutter estimation based on adaptive Gaussian mixture model (GMM). Firstly, the adaptive GMM is utilized to fit the spatial distribution of the clutter, and the clutter intensity is estimated. Then, the clutter intensity is applied to the MHT method to effectively improve the accuracy of track score calculation and optimal hypothesis track estimation. The simulation results demonstrate that the proposed method obtains better data association accuracy and track maintenance performance than the standard MHT algorithm and the MHT-GMM algorithm in an unknown heterogeneous clutter observation scene.