Expectation Maximization Algorithm for Multi-Feature Aided Probabilistic Data Association Based on Least Square

A multi-feature aided probabilistic data association is developed in this paper. In the environment of high noise and high-density clutter, tracking multi-target used standard PDA algorithm may occur a poor association accuracy. Aiming at this problem, we incorporate target feature information into the tracking process, and use the expectation maximization (EM) algorithm to compute the least square’s error function of estimating target states iteratively. Via gradually modifying the estimated parameters that are target states and multi-feature aided associated probabilities, the target states could be estimated more accurately. The simulated results show that the new algorithm can strengthen the discrimination between targets and clutter, reduce the missed tracking or false tracking due to similar feature information, weaken the dependence on the probability of detection, and improve the estimated accuracy of target states significantly.