Maximum correntropy quadrature Kalman filter based interacting multiple model approach for maneuvering target tracking
Bao Liu,Ziwei Wu
DOI: https://doi.org/10.1007/s11760-024-03642-y
IF: 1.583
2024-12-07
Signal Image and Video Processing
Abstract:This paper proposes a Maximum Correntropy Quadrature Kalman Filter Based Interacting Multiple Model approach (IMM-MCQKF) for the state estimation with the non-Gaussian noise and model uncertainty problem in the maneuvering target tracking. The performance of Quadrature Kalman Filter (QKF) will significantly decrease when the noise is non-Gaussian, especially under the heavy tailed noise. The proposed IMM-MCQKF approach not only solves this problem, but also ensures a good tracking accuracy with the mode switching caused by the target maneuvering. Specifically, a Maximum Correntropy Quadrature Kalman Filter (MCQKF) combines the maximum correntropy criterion and statistical linearization regression model into the QKF algorithm to increase the estimation accuracy under the non-Gaussian noise. Moreover, we introduce a fixed-point iteration method to ensure that the MCQKF algorithm has a closed solution. Secondly, the MCQKF is embedded into the interacting multiple model framework to address the problem of mode switching caused by target maneuvering. Finally, the simulation results demonstrate the tracking effectiveness of the IMM-MCQKF in maneuvering target tracking scenarios with non-Gaussian noise compared with IMM-EKF, IMM-UKF, IMM-CKF, IMM-QKF, AIMM-UKF, and IMM-AMTP algorithms.
engineering, electrical & electronic,imaging science & photographic technology