Before The Weak Target Detection Based On Particle Filter Tracking Algorithm Research

In the modem electronic warfare, the radar is faced with a lot of low echo energy targets, such as stealth aircraft and cruise missile. The traditional detect before track (DBT) technology detected by a single frame, is very easy to cause the leakage alarm; track-before-detect (TBD) technology directly using the original data as input, detected after multi-frames accumulation, can effectively detect and tracking dim target in the complex background. Particle filter (PF) is a kind of recursive track-before-detect approach, a large number of particles is used for the approximate of target state posterior probability density, which has a unique advantage in a non-linear, non-Gauss tracking problem.The dissertation focuses on the problem of weak target detection and studies the particle Filter based TBD (PF-TBD) technique. It mainly comprises:1. The TBD and DBT techniques are compared,the merits and demerits of a variety of TBD technology are analysis. Expound the causes why this paper uses PF-TBD technology to study weak target detection.2. Introduces the particle filtering technology to form, established a basic TBD model, and the theory of particle filter to realize track before detection is derived, which are the basis of the PF-TBD algorithm.3. Two kinds of mainstream PF-TBD algorithm, namely the Standard PF-TBD (SPF-TBD) algorithm, and Efficient PF-TBD (EPF-TBD) algorithm are compared. Simulation results show that, EPF-TBD has better detection performance and tracking accuracy than SPF-TBD.4. Aiming at the problem of particle degeneration and sample poor, The Genetic Resampling based PF-TBD (GR-EPF-TBD) algorithm is proposed, which can increase the diversity of samples, and has better detection performance and tracking accuracy than SASIR-TBD.5. Aiming at the problem that there are possibly a number of targets in the observation area. The multiple targets PF-TBD (Multi-SPF-TBD) algorithm when the number of the target is known is proposed, which can achieve multiple target detection.