Research On Algorithms Of Multi Targets Tracking Under Unknown Measurement Noise

Multi target tracking technology is an important issue in the field of information processing. It has a very broad application prospect in the military and civil fields. More and more domestic and foreign scholars pay close attention to it. Recently, because the multi-target tracking method which is based on random finite sets does not need to solve the data association problem, it brings a new energy to the multiple target tracking area.In this thesis, we deal with the problem of multi objects tracking based on the random finite set under the circumstance of unknown parameters. This thesis mainly studies the point targets tracking with the unknown detection probability and unknown measurement noise covariance. And also studies the extended targets tracking under the condition of unknown measurement noise covariance.Besides studies the point or extended targets tracking with glint noise. The main research contents of this thesis are as follows:On the basis of the traditional algorithm, a novel algorithm based on Variational Bayesian cardinalized probability hypothesis density(VBCPHD) is proposed to solve the problem of point target tracking with the detection probability and measurement noise covariance being jointly unknown. At the basis of traditional detection probability estimation method, the CPHD method based on VB is used to iteratively estimate the target state and measurement noise covariance. Simulation results show the excellent target tracking performance.For multi-target tracking problems, the Gaussian distribution has commonly been used for representing the measurement noise statistics due to its mathematical simplicity and effectiveness. But this often leads to the poor target tracking performance, especially when the measurement noise is the glint noise. In this thesis, the glint noise is modeled as Student's t distribution. And the VB method is used to iteratively estimate the parameters of target state, freedom and measurement noise inverse covariance. Experiments results show that the VB is proper to be used to track the point targets under the condition of glint noise.A novel CPHD algorithm under the assumption of a glint measurement noise model with unknown inverse covariance. Noise parameters are assumed to have a Gamma prior distribution so that the predicted and updated PHDs can have mixture of Gaussians representations. A Variational Bayesian Expectation Maximization(VBEM) procedure is applied to iteratively estimate the parameters of the mixture distributions through CPHD prediction and update steps. Simulation results show that the proposed algorithm can adaptively track extended targets under conditions of unknown target number and glint measurement noise, while achieving higher precision compared against the traditional approach. However the proposed method costs a longer time.