Font Size: a A A

Research On Track-before-detect Algorithm For Dim Targets Based On Particle Filter And Its Implementation On FPGA

Posted on:2019-01-09Degree:MasterType:Thesis
Country:ChinaCandidate:Z LiuFull Text:PDF
GTID:2428330566474315Subject:Signal and Information Processing
Abstract/Summary:PDF Full Text Request
As stealth technology continues to improve,the radar cross section of the military target is getting smaller and smaller and the intensity of echo signal received by radar becomes weaker and weaker,with the result that detection and tracking of targets become more and more difficult.In the complicated environment where the external interference is very serious,the intensity of echo signal of the target is submerged in the noise easily,which also poses a great challenge to the detection and tracking.It is difficult to find the target effectively under the condition of low signal-to-noise ratio using the traditional tracking-after-detection(TAD)method,and it is easy to detect the dim target effectively using tracking-before-detection(TBD)by accumulating the faint signal of multiple frames.In this paper,the following research is carried out on the problem with tracking-before-detection of dim targets;1.Through the flow chart of the algorithm,the characteristics and differences of the detection-after-tracking(TAD)and tracking-before-detection(TBD)algorithm are analyzed and compared clearly,and the conclusion that TBD is superior to TAD is drawn in the case of problem with detection and tracking of dim target.2.The principle of generic particle filtering is discussed in detail,and it is found that resampling and importance function are the crux of affecting particle filter performance.The principle and performance of the standard particle filter based tracking-before-detection algorithm(SPF-TBD)proposed by Salmond etc.and the Rutten particle filter based tracking-before-detection(Rutten PF-TBD,RPF-TBD)proposed by Rutten etc.are analyzed and compared,and finally the latter one we prefer is the PF-TBD method for this paper.3.The principle and characteristics of Gaussian sum particle filter(GSPF)approximating posterior distribution with sum of several weighted Gaussian and free of resampling is analyzed.With the GSPF and RPF-TBD combined,the Gaussian sum particle filter based tracking-before-detection(RGSPF-TBD)algorithm is proposed.Using quasi-Monte-Carlo(QMC)random number to replace pseudo-random number in GSPF,a quasi-Monte-Carlo Gaussian sum particle filter(QMC-GSPF)algorithm is proposed.With QMC-GSPF and TBD combined,QMC Gaussian sum particle filter based track-before-detection algorithm(QMC-RGSPF-TBD)is proposed.The simulation results show that the two TBD algorithms presented in this paper have good performance.4.The module structure of PF-TBD is studied,and the realization scheme of PF-TBD on FPGA is fulfilled with referring to hardware realization of particle filter,which fills the gap in the field.
Keywords/Search Tags:Particle Filter, Track-before-detection, Quasi-Monte Carlo Sampling, Free of Resampling, FPGA
PDF Full Text Request
Related items