A Maneuvering Extended Target Tracking IMM Algorithm Based on Second-Order EKF

Aiming at the problem of poor tracking performance of traditional filtering algorithms for maneuvering extended targets, an improved interactive multimodel second-order extended Kalman filter (IMM-SOEKF) algorithm is proposed in this article. First, the Markov transfer probability matrix is updated using the probabilities within the neighboring time points between each model to improve the switching speed of the models in the algorithm and the matching accuracy. In addition, according to the state parameters and shape information of the target, the second-order extended Kalman filter (SOEKF) is used for tracking the estimation of the target. By incorporating the measurement covariance equation from the filtering algorithm into the likelihood function calculation, new likelihood probability and model weight assignments are obtained using the maximum likelihood function method to improve the tracking accuracy for extended targets and the robustness of the algorithm. Finally, simulation and data processing results show that the algorithm has higher tracking accuracy compared with other state-of-the-art algorithms.