Improved Track-before-detect Algorithm Based On Random Finite Set

When the target's echo signal is weak,it is very difficult to detect and track the weak target by using the traditional method of detect-before-track.With the continuous development of Random Finite Set(RFS),the target tracking algorithm based on RFS has gradually developed into an important topic,and different filtering algorithms have emerged.In this algorithms,Among them,Probability Hypothesis Density(PHD)filtering is a new algorithm proposed in recent years.The RFS theory is used in the Track-Before-Detect(TBD)algorithm to improve the detection and tracking performance of weak targets.Supported by RFS theory,and based on an deep understanding of the PHD,this paper focuses on PHD smoothing algorithm,pruning and merging algorithm for gaussian mixing component and the TBD algorithms for multiple weak targets.The main works and achievements of this thesis are as follows:(1)Research on smoothing algorithm based on PHD filtering.Focus on the closed solution of the forward-backward PHD smoother under linear Gaussian condition:Gaussian Mixture Probability Hypothesis Density(GM-PHD)smoothing algorithm.In this paper,an improved GM-PHD multi-target smoothing algorithm is proposed for the misestimation problems caused by the disappeared targets.The MATLAB simulation experiment proves that the improved GM-PHD smoothing algorithm can effectively solve the problem of abnormal smooth estimation caused by the disappearance of the target.(2)Research on pruning and merging algorithm based on PHD filtering.Focus on the pruning and merging steps of the GM-PHD algorithm.The traditional trimming and merging algorithm does not fully consider the parameter information of the Gaussian component,and the error merges when the target is moving at a close distance,resulting in low tracking accuracy.In order to solve this problem,this paper proposed an improved merging algorithm.The improved algorithm fully considers the three important parameters of Gaussian components and corrects the merging formula of Gaussian components.The MATLAB simulation experiment proves that compared with the traditional pruning and merging algorithm,the improved pruning and merging algorithm can accurately estimate the target state at the moment when the target disappears,and has better target tracking performance.(3)Research on tracking before detection algorithm based on PHD Filter.The TBD algorithm of Gaussian particle PHD filter is mainly studied.Combined the improved pruning merging algorithm with TBD problem to handle complex scenes of multiple target adjacent and cross motions.Finally,two different simulation scenarios are set up,including target adjacent and cross.The MATLAB simulation experiment proves that the improved algorithm can better deal with target adjacent and cross motion with better signal-to-noise ratio and has good detection and tracking performance.