Research On Probability Hypothesis Density Multi Target Tracking Algorithm Based On Random Finite Set

With the continuous progress of electronic communication technology,intelligent networking technology and environmental sensing technology,multi-target tracking is closely concerned by domestic and foreign research institutions and high-tech enterprises,and plays an important role in many engineering fields.Data association is widely used in multi-target tracking algorithm,but the complexity of data association is high.The scholars represented by Wax introduced random finite set into multi-target tracking technology,and used set to describe the relationship between target and measurement,avoiding the process of data association,which has a wide application prospect.When the new target information is uncertain,the target is missed,and multiple targets are moving in parallel or interleaved,the filtering performance of the tracking algorithm based on random finite set is often greatly affected.This paper proposes corresponding solutions to these problems,which lays a foundation for improving the tracking performance and practicability of the algorithm.Aiming at the problem of unknown strength of new target,an improved multi-target tracking algorithm based on measurement driven is proposed.The improved algorithm firstly removes the clutter and predicts the new target strength according to the velocity characteristics;then divides the measurement set based on the measurement segmentation strategy,and then selects the measurement set according to the target category to update the new target strength,which effectively avoids the interference of other categories of measurement and clutter when updating the target strength.The simulation results show that the improved algorithm has lower computational complexity and more accurate prediction of target state and number.A GM-PHD filter based on target state extraction is proposed to solve the problem of true and false targets missing detection in the test area.The improved algorithm deals with the problem of true and false missing detection of the target properly through the target state extraction scheme and the target weight update scheme.The simulation results show that the filtering performance of the improved algorithm is stable and the tracking performance is good.In order to solve the problem that the weight of target components cannot be assigned correctly due to the multi-target adjacent state,this paper proposes an improved GM-PHD filter based on the weight redistribution of target components.By establishing the normalized weight matrix of the target,the improved scheme can find and adjust the adjacent situation of the target in time,and prevent the wrong combination of the target components by improving the component reduction and combination scheme.Finally,MATLAB is used to simulate and compare the improved algorithm with other filtering algorithms,which verifies that the improved algorithm has higher accuracy and less computation in multi-target tracking.