The Multi-sensor Multi-target Track To Track Fusion Algorithm

As the modern battlefield environment changes significantly, targeted high mobility, high clutter of the battlefield environment, low detection rate, high false alarm rate and other issues, making the design of multiple targets track system more difficult. This article is based on the actual background of the project, a number of low cost radars component distributed network detection system for multiple targets tracking.It requires research and design for the structure and modeling of the detection system. By in-depth analysis of the theory and technique of multisensor information fusion, this paper studies the multi-target tracking in five aspects including track initialization, data association, maneuvering target state estimation, track association and track fusion.Firstly, a new algorithm is proposed to solve the problem of quick target tracking initialization in clutter. The advantages of a new track initial model and one-step delay approach are integrated. An intermediate stage is added to the current common track initiation model. The one-step delay approach is used in processing the intermediate track. By this algorithm, the design flexibility is improved and that probability of false tracks is reduced.Secondly, to implement the problem of multiple maneuvering targets data association and track maintenance in cluttered environment, a kind of algorithms interactive multiple model joint probabilistic data association (IMM-JPDA) has been researched and implemented. The interacting multiple model (IMM) algorithm is a popular algorithm for Multi-target Tracking. But the performance of IMM algorithm will be greatly reduced under the dense cluttered environment. JPDA is usually used to solve the problem of tracking multiple targets under the environment of heavy clutter. Therefore, we can combine two algorithms to produce a new IMM-JPDA algorithm. While it accesses to the advantages of the previous two algorithms, the IMM-JPDA algorithm inherits their weakness which is a huge amount of calculation. In this paper, a new IMM-JPDA algorithm is proposed. This algorithm will decrease the computational burden and be of much better performance.Thirdly, the paper analyzes and compares the distributed track association algorithms which are currently widespread adoption. Include weighted and modified, independent sequential track association algorithm, nearest neighbor and K-nearest neighbor track association algorithm,fuzzy track association with dual thresholds, gray track association algorithm.Compares in different simulation environment, the advantages and disadvantages of the various algorithms. Finally, study on current domestic and international wide fusion algorithm, such as simple convex combination, Bar Shalom-Campo, no feedback fusion algorithm, selecting the best relative performance the no-feedback fusion algorithm simulation analysis. Simulation results show the feasibility of this algorithm, multiple low-cost small radar track fusion, effectively improves the tracking precision of the system.