Maneuvering Target Joint Detection and Tracking Using Multi-Frame Integration

The authors present an efficient joint detection and tracking algorithm for maneuvering target. To exploit the space-time correlation of the target among successive frames, we derive an expectation multi-frame test statistic containing a quantifiable term of target motion character by probabilistic dynamic programming. A joint filtering and smoothing estimation based on the interacting multiple model (IMM) is developed to design and calculate the state transition probability of target motion. The proposed algorithm promises a satisfying detection performance in that the designed state transition probability can steer the test statistic to be integrated along the real target trajectory more likely. Besides, better tracking performance can be obtained since the outputs of the proposed algorithm are filtered or even smoothed trajectories of the detected targets. Simulation results illustrate the performance of the proposed algorithm.