A Particle Filter Tracking Algorithm of Multi-features Fusion Based on Energy Cumulant

A particle filter tracking algorithm of multi-features fusion based on energy cumulant is proposed in this paper. This algorithm mainly focuses on the dim target tracking problem under complex background of infrared image sequence, and analyzes the different features of infrared dim targets. Since the particle filtering algorithm gives the advantage of multi-features fusion, this paper combines the four features, such as gray scale value, local entropy feature, local energy feature and high-frequency histogram feature, to calculate the particle weights which greatly improves the tracking accuracy, and uses energy cumulant algorithm to suppress the background and improve the signal to clutter ratio (SCR). The experimental results on both synthetic and real-world data demonstrate that, the proposed algorithm has substantial improvements in terms of tracking accuracy and robustness over the traditional particle filtering algorithms.