Research On Algorithms Of Joint Tracking And Classification For Maneuvering Extended Target

With increased resolution of radar and infrared sensors, extended object tracking has received widely attention by domestic and foreign scholars. In recent years, the research on algorithms of joint tracking and classification for extended object using random matrix is concerned because of the simple computation and the ability to track the target. Using the Bayesian filtering and the random matrix as the theoretical basis, the thesis aims at two kinds of ellipse and non-ellipse extended targets to study the approach of joint tracking and classification for maneuvering extended target, which is of great important theoretical significance and application value. The main research contents of this thesis are as follows:1. The basic theory of maneuvering extended target tracking is studied, which includes the object dynamics model, the Bayesian filtering theory, the random matrix algorithm and the multiple-model algorithm. Firstly, the multiple-model algorithm for maneuvering extended target tracking is studied based on the conventional dynamics model. Secondly, both target tracking state estimation and the algorithm of target tracking using random matrix are studied based on the theory of Bayesian filtering. Finally, the algorithm of target tracking using random matrix is studied in order to estimate the kinematic states and the extensions of a maneuvering extended target simultaneously.2. Combined with the interacting multiple-model approach, an algorithm of joint tracking and classification of maneuvering ellipse extended object using random matrix is proposed for maneuvering ellipse extended object. Combined with the prior information of target size, shape, direction and so on, the algorithm takes the Bayesian filtering as the theoretical framework, and uses the random matrix to estimate the target's state while the multiple-model approach is used to deal with the maneuvering problem of ellipse extended object. The simulation experiment shows that the proposed algorithm can estimate the target's kinematic state, the extension and the classified state in real time, and achieve joint tracking and classification of maneuvering ellipse extended object.3. Combined with the interacting multiple-model approach, an algorithm of joint tracking and classification of maneuvering non-ellipse extended object using random matrix is proposed for maneuvering non-ellipse extended object. The approach assume that a non-ellipse extended object can be approximated by a number of mutually independent sub-ellipses, each described by a random matrix, and the relations between the sub-ellipse and the main ellipse are represented by the structural information. Then the state of the main ellipse is estimated by the proposed algorithm of joint tracking and classification of maneuvering ellipse extended object using random matrix. Every sub-ellipse can be estimated by utilizing the structural information, and joint tracking and classification of maneuvering non-ellipse extended object is realized. The simulation experiments show that the proposed algorithm can improve the information loss caused by elliptic approximation and estimate the target's kinematic state, the extension and the classified state in real time.