Track Before Detect Algorithm For Infrared Dim Target Based On Particle Filter

With the development of missilery and infrared techniques, infrared imaging guidance technique is paid more and more attention to, and the technique of dim infrared target detection and tracking has become one of the key techniques in the infrared imaging system. At present, there are two main techniques to detect and tract dim targets in infrared image sequences. One is detect before tract (DBT), and the other is tract before detect (TBD). DBT has good performances of detection and tracking when it is applied to dim target in high signal-to-noise ratio (SNR) environments. But in very low SNR environments, TBD has unique advantages. In the dissertation, detection and tracking dim target in low SNR environments is researched and TBD algorithm based on particle filter (PF) is used.In the dissertation, the basic theory of particle filter and its improved algorithm are presented, and simulation results show that particle filter is suitable to nonlinear and non-Gaussian problems and its performance is superior to extended Kalman filter. And then TBD algorithm based on particle filter is researched. This algorithm utilizes motion equation, observation equation and raw data of sensor, obtains posteriori probability distribution of target through particle filter and takes target presence probability as detection criterion. Estimate target's position in the plane if it is detected. Finally the algorithm is applied into real infrared image sequences. Because there is much blackground noise in infrared images, grayscale morphology filtering is used to inhibit blackground noise. And then the algorithm is applied to detect and track dim target. Experimental results verify that the algorithm can successfully detect and track dim target.