A Laplace-based particle filter for track-before-detect

Track-before-detect (TBD) is an efficient approach for tracking a target with low signal-to-noise-ratio. Unlike classical detection methods, TBD does not require thresholding. TBD integrates the signal of the target over space and time. In this paper, a particle filter is proposed, where the proposal density is constructed thanks to local Laplace approximations at the modes of the posterior density. This particle filter is tested on a simulated TBD problem, where the observations are a sequence of infrared images.