Title :
Target maneuver detection using a particle filter with spawn model and particle labeling
Author_Institution :
Dept. Sensor Data & Inf. Fusion, Fraunhofer FKIE, Wachtberg, Germany
Abstract :
This paper presents a novel single target particle filter with spawn model and particle labeling approach, abbreviated SL-PF. The purpose of this filter is to detect instantaneously occurring target maneuvers, e.g. course changes of maritime targets, and to provide accurate tracking performance before and after the maneuvers. The key idea is to borrow the spawn model from the probability hypothesis density (PHD) filter since this model is naturally suited for these kinds of maneuvers. Secondly, each particle in the filter carries a label which is updated in a systematic manner in the spawning step so that it is possible to recognize spawned particles representing a target maneuver. This provides an integrated maneuver detection procedure within the particle filter. Monte Carlo simulations verify the SL-PF approach and indicate a significant estimation accuracy improvement compared to a conventional particle filter.
Keywords :
Monte Carlo methods; object detection; particle filtering (numerical methods); Monte Carlo simulations; PHD filter; abbreviated SL-PF; particle filter; particle labeling; probability hypothesis density filter; spawn model; target maneuver detection; tracking performance; Atmospheric measurements; Covariance matrix; Labeling; Monte Carlo methods; Noise; Particle measurements; Target tracking; labeled particles; maneuver onset detection; maneuvering target; particle filter;
Conference_Titel :
Sensor Data Fusion: Trends, Solutions, Applications (SDF), 2012 Workshop on
Conference_Location :
Bonn
Print_ISBN :
978-1-4673-3010-7
DOI :
10.1109/SDF.2012.6327915