Research On Track-before-detect Algorithms Based On Particle Filter For Multiple Dim Targets

Due to the stealth technology used in modern warfare or small size of itself andother factors, using traditional radar and other sensors to detect these weak targetswhose echo signal energy is low,are more difficult. Tracking BeforeDetection(TBD),which introduces trace algorithm before detection,take full advantageof the target echo signal information use the original multi-frame signal energyaccumulation, improve the performance of detection and track weak target under thestrong clutter. Since the targets are usually in the form of multi-batch or cluster in theactual battlefield environment, accurately estimating the number of targets and thestate of each target, and forming the correct battlefield,are of great importance in thetheoretical significance and military application value, and become the hot point inresearch on radar probing test today. This paper does research on integrationtechnologies for multiple weak targets detection and tracking in complex environmentunder the pre-research project support. The main research contents are as follows:1. For detecting and tracking individual weak target, TBD algorithm based onparticle filter (PF) is studied. Showing the posterior probability density with theweights of the particle collection can achieve weak target detection and tracking.Through the analysis of performance indicators based on PF-TBD algorithm suchas probability of detection, distance estimation, etc., the situation of lowsignal-to-noise ratio PF-TBD algorithm can well detect the presence of a target,and the distance estimation performance is well.2. For detecting and tracking multiple weak targets whose the number of targets areknown, TBD algorithm based on particle filter for multiple weak targets isstudied. Introducing a subject Markov model transfer process variables candescribe the case of the current number targets, and using hypothesis judgmentwhether target presence or absence, can estimate the number of the targets.Showing the posterior probability density with the weights of the particlecollection can estimate the state of each target and achieve multiple weak targetsdetection and tracking. Finally this paper analyzes the algorithm of targetdetection and tracking performance under different SNR, validates that thealgorithm can effectively estimate the number of targets accurately, and has a good tracking performance in low SNR conditions.3. For detecting and tracking multiple weak targets whose the number of targets areunknown, TBD algorithm based on Probability Hypothesis Density (PHD) formultiple weak targets is studied. By dealing with multi-objective posteriorprobability density of first moment, it can recursively pass the target statedistribution of information, use the theory of random set without needingcomplex data correlation e to extract the number of the targets and status, thenextract the state to determine its trajectory through clustering analysis.Combining PHD with TBD can solve the multiple weak targets under low SNR’sdetecting and tracking problems. Through simulation PHD-TBD is validated inthe case of unknown target number and can accurately estimate the target number,and has a good tracking performance.