Multiple Dim Moving Targets Detection based on Bayesian Filtering

This paper uses the multi-spectral information and temporal information of target to estimate the state of multiple moving targets, and proposes a method for detecting multiple dim moving targets based on Bayesian filtering. The research idea is as follows: A dynamic model and an image measurement model of multi-target are established. The Markov chain is used to model the target existence variables. The problem can be described as a hybrid filtering problem involving discrete and continuous variables. The solution of hybrid filtering is derived using Bayesian theory. Based on this, multi-spectral infrared image sequence is used to derive a multi-target detection algorithm. The proposed multiple interactive hybrid particle filter is achieved. Simulation scenarios are used to verify the effectiveness of the proposed algorithm.