Multi-target Tracking Algorithm And Implementation Of Research

With the rapid development of science and technology, especially IT technology, the target-tracking environment has changed significantly. The technology of target tracking is faced with a serious challenge of strong maneuvering, high clutter, low detection probability and high false-alarm ratio. Some of the key problems in the field of maneuvering target tracking in clutter environment are studied. The main contributions are as follows.1. Modeling Maneuvering Target in Cartesian Coordinates — Base on the study of time-correlated model, we propose a new Adaptive Constant Acceleration (ACA) model. Comparing with the non-zero mean time-correlation model, the ACA is proved equivalent to the non-Zero mean time-correlation in filter accuracy while much computational load is saved.2. Modeling Maneuvering Target in Polar Coordinates - The dynamic model of a maneuvering target can be easily transformed from Cartesian into polar coordinates. One important merit of such transformation is in that the dynamic errors in different coordinates can be decoupled. Thus the computation burden can be greatly decreased. Simulation results show that the algorithm has lower computational burden and same filter accuracy, compared with the traditional method.3. Multiple Model Algorithm & Data Association - In order to overcome the defects in construction and theory of Interacting Multiple Model Probabilistic Data Association (IMMPDA) algorithm proposed by Bar-Shalom and Blom, we propose a new Comprehensive IMMPDA/IMMJPDA algorithm—C-IMMPDA/C-IMMJPDA algorithm. Simulation results show that this algorithm will decrease the computational burden and be of much better performance.4. Track Initialization - A new algorithm for tracking filter initiation is presented. The a priori knowledge of velocity is used for constructing the gate of filter. The algorithm is illustrated via Monte Carlo simulation where a significant improvement over two other algorithms is achieved.5. Heterogeneous Information Fusion - As we known, ground targets are always constrained by roads or rivers. The information about roads or rivers aroundthe target can be provided by electronic map. Hence through fusing electronic map and traditional sensor position measurements, a satisfactory heterogeneous fusion method for a Track While Scan (TWS) surveillance-system is presented, where the estimation of tracking filter based on traditional sensor position measurements is further modified by a hypothesis testing.6. Design of Tracking System and Software implementation- We present a whole design plan for multiple maneuvering targets tracking system, and design a multiple maneuvering targets tracking simulation software package. By using object-oriented programming and hierarchical structure, it is much clear in layer and easy to use and maintain, and provides an effective simulation environment for further research work.