TY - GEN
T1 - A robust filter based on Pearson type VII mixture distribution
AU - Cui, Yujin
AU - Zhao, Huachuan
AU - Wang, Guochen
N1 - Publisher Copyright:
©2025 IEEE.
PY - 2025
Y1 - 2025
N2 - This paper addresses the single-target tracking problem in the presence of complex non-Gaussian noise. We propose a Variable Kurtosis Mixture Model (VKMM) filter based on variational Bayesian inference, which uses the Pearson type VII (PTVII) distribution mixtures to model unknown noise. An improved variational Bayesian method is developed to jointly optimise model parameters and system states, and the filter incorporates a matrix decomposition strategy to integrate prior information and multi-source measurements within a unified framework. Monte Carlo simulation results indicate that this method outperforms the comparison algorithms in both estimation accuracy and robustness.
AB - This paper addresses the single-target tracking problem in the presence of complex non-Gaussian noise. We propose a Variable Kurtosis Mixture Model (VKMM) filter based on variational Bayesian inference, which uses the Pearson type VII (PTVII) distribution mixtures to model unknown noise. An improved variational Bayesian method is developed to jointly optimise model parameters and system states, and the filter incorporates a matrix decomposition strategy to integrate prior information and multi-source measurements within a unified framework. Monte Carlo simulation results indicate that this method outperforms the comparison algorithms in both estimation accuracy and robustness.
KW - Cubature kalman filter (CKF)
KW - Non-Gaussian noise
KW - Pearson type VII distribution
KW - variational Bayesian (VB)
UR - https://www.scopus.com/pages/publications/105046977565
U2 - 10.1109/EISP70220.2025.00009
DO - 10.1109/EISP70220.2025.00009
M3 - 会议稿件
AN - SCOPUS:105046977565
T3 - Proceedings - 2025 2nd International Academic Symposium on Electronic Information and Signal Processing, EISP 2025
SP - 11
EP - 16
BT - Proceedings - 2025 2nd International Academic Symposium on Electronic Information and Signal Processing, EISP 2025
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2nd International Academic Symposium on Electronic Information and Signal Processing, EISP 2025
Y2 - 28 November 2025 through 30 November 2025
ER -