Research On Multiple Maneuvering Targets Tracking

Due to its extensive application perspective in military and civil industries, study on target tracking has been paid more attention by researchers in the world. In recent years, target tracking technology has been largely studied, and plentiful results have been acquired. Multiple maneuvering targets tracking is a research focus of the field of target tracking at present. It classifies the measurements of numerous sensors according to their sources, gains the tracks of every target, and then analyzes the veracity and reliability of the target tracks. This paper is concerned with two aspects of multiple maneuvering targets tracking technology, those are data association and tracking maintenance.Firstly, several popular maneuvering target tracking algorithms, such as Kalman filtering algorithm, extend Kalman filtering algorithm,"current"statistical model algorithm and interacting multiple model algorithm are studied.Then,a modified"current"statistical model algorithm based on fuzzy logic is presented. The simulation results demonstrate the availability of these tracking algorithms.Secondly, methods of tracking initiation and termination of multiple targets are studied. Then several methods of data association are studied in details and a new Modified Current Model-Probability Data Association algorithm is presented. The simulation results demonstrate it outperforms Interacting Multiple Model-Probability Data Association(IMM-PDA)algorithm in accuracy.Thirdly, by reason of many nonlinear models in target tracking, three kinds of nonlinear filtering algorithms are studied. These algorithms include extend Kalman filtering algorithm, unscented Kalman filtering algorithm, and particle filtering algorithm. The merits and demerit of these algorithms are analyzed and compared.Finally, a new data association algorithm based on support vector machine for multi-target is presented. The filtering measurement innovation is considered to be the input of the support vector machine for judging the association between each target measurement and track. Then it uses a method integrated of interacting multiple model and unscented Kalman filtering algorithm to study the tracking problem of maneuvering targets.The theory of multiple maneuvering targets tracking is discussed in this thesis. Adaptive filtering algorithms and data association algorithms are studied. Unscented Kalman filter and support vector machine are introduced to the study of target tracking. The simulations show the availability of the algorithm. The future develop direction of multi-target tracking is also presented in this thesis.