Font Size: a A A

Complex Waters High Density Multi-Target Tracking Methods Research

Posted on:2017-01-17Degree:MasterType:Thesis
Country:ChinaCandidate:R H HuFull Text:PDF
GTID:2272330482478439Subject:Information and Communication Engineering
Abstract/Summary:PDF Full Text Request
With the development of the shipping industry, Waterway transport situation is getting complex and inland waters vessel traffic density has increased annually. These give rise to safe navigation problems. But the tracking volume and tracking reliability of existing VTS system can not fully meet the tracking requirement under complex waters and high density target environments. This brings difficulties for the water traffic safety management. Therefore, it is necessary to improve the multi-target tracking capability of VTS system to improve overall system performance.This paper detailedly analyze and compare the existing mature multi-target tracking algorithm theory, try to select some methods that possess high tracking reliability and high project realize value simultaneously. And then improve the selected methods under the application background, form a set of complete muti-target tracking solution used in the actual project.Filtering algorithm, association algorithm, multiple model algorithm is the theoretical basis of the multi-target tracking. This paper detailed discussed and design these algorithms respectively. For high-density target application background, this paper compared the application scene and relative merits of Nearest Neighbor Data Association (NNDA), Point-Track Association (PTA), Probabilistic Data Association (PDA), Joint Probabilistic Data Association (JPDA) and the newest Probability Hypothesis Density (PHD) algorithms. At last, unite the advantage of PTA and PDA and introduced characteristic difference value of the target echo size to improve the association algorithm. Proposed a data association method in high density target situation. For the limitation that PHD method need prior knew the tensity of birth target, this paper propose a unknown birth target intensity PHD filter based on phased track management for instance of Gaussian Mixture (GM) solution.Under complex water, target information may appear close overlapping, shielding, false echoes interference. These problems seriously limit the reliability and stability of the tracking system. This paper gives solutions respectively. At last, combine with track initiation and track management to design a compete real-time radar multi-target tracking system.
Keywords/Search Tags:Multi-target tracking, Joint Probabilistic Data Association, Probability Hypothesis Density, Gaussian Mixture, False Echoes Processing
PDF Full Text Request
Related items