| The basic function of radar is to transmit ,to receive the specific signals and to discover objects and to measure their coordinates. The first aim of radar signal processing is to minimize the disturbance of noise and clutter so as to obtain necessary information and improve the quality of the communication. Therefore, the research on radar signal processing method has great importance in both theory and application.The theories and background of EM algorithm are discussed firstly. And the material formula is deduced through analyzing MLE. At the same time, the illustration of dealing with multiple Gaussian distribution is brought forward. Hereby, it is concluded that the super complex data of multi-dimensional multiple normal distribution is dealt with by EM algorithm. Furthermore, the precision is improved with increasing the number of samples. And then, according to the material example, the model of Markov and Kalman filter are given out. It is concluded that the Markov model could simulate the data of radar measured in clutter very much, by Viterbi algorithm. Afterwards, the localization of the Kalman filter model is exampled.The application of EM algorithm in multi-target tracking is the main concerning problem of this thesis. Firstly, the maximum posterior (MAP) model for the state estimation is built on the basis of EM algorithm. Then, the maximum procedure in EM algorithm is accomplished by means of the discrete optimization technique. Finally, as a result of the procedure, the maneuver input sequence is determined optimally, and the measurement sequence from targets is separated efficiently. Consequently, the more precise state estimates of the target can be obtained. The presented algorithm solved efficiently the incomplete data problem of state estimation in MAP sense. The simulation results indicate that the presented algorithm behaves more superiorly than the interacting multiple model-probabilistic data association. |
PDF Full Text Request