Track Segment Association of Maneuvering Target Based on Expectation Maximization

Abstract

Because of the high maneuverability of targets, track breakages are common in the tracking process. With the purpose of stitching the track segments and achieving better tracking results, we adopt an algorithm to stitch the track segments based on expectation maximization (EM). The EM algorithm can be used to estimate and identify the maneuvering targets' state and angular velocities simultaneously. It consists of two steps. The expectation (E) step is implemented by an extended Kalman filter (EKF) and extended Rauch-Tung-Striebel smoother (ERTSS). The maximization (M) step is implemented by genetic algorithm, which can achieve the Maximum likelihood sequence estimation for unknown parameters. Experiments show that this algorithm can achieve better tracking results. Moreover, it also exhibits good capability when estimating the unknown parameter.