Pmht Algorithm And Its Application In The Days Of Radar Target Tracking

Data association plays very important role in multi-target tracking. And the traditional data association algorithms need enumerating all of the possible associations between measurements and targets, therefore, the computational burden will increase exponentially with the increasing number of targets and observations. In this thesis, the probabilistic multi-hypothesis tracker (PMHT) based algorithms are studied and applied to conventional or unconventional target tracking. The main contributions are as follows:1. We summarize the latest theory and application works in this field. Based on analyzing theoretical derivation formula, the result shows the problems of the probability multi-hypothesis track (PMHT) algorithm that exist in the theory and application.2. Based on analyzing of the canonical PMHT algorithm, as well as various improved algorithms of the PMHT deeply, we show the weaknesses of each algorithm. other,the algoritons Simulation and comparison analysis take on some availableinformation and conclusions.3. According to the iterating and batch processing of the PMHT, A fuzzy adaptive probabilistic multi-hypothesis tracking (FA-PMHT) algorithm for maneuvering targets is proposed in this paper , which extends the traditional optimization of data association of PMHT to the joint optimization of data association and dynamic modeling through introducing a set of fuzzy rules, variance of process noise in dynamic model is adaptively and iteratively optimized. Hence our FA-PMHT obtains better adaptive ability no matter that the target maneuvers or not. Computer simulation of a benchmark target tracking shows that our FA-PMHT is superior in tracking precision and meets real time requirement.4. PMHT algorithm is based on the Bayesian theory, so it has the characteristics of easy to expand the framework structure. For this , A adaptive Homothetic-PMHT algorithm based on image information is proposed in this paper. This algorithm utilizes the integrated attribute and position information . First, according to the fuzzying processing for the attribute measurements which we obtained from the image, we get the probability information of the target attribute measurements. Then integrating the probability information of the target attribute and measurements...