A Parallel Centralized Multi-sensor General Association Algorithm

This paper has researched on the problem of centralized multi-sensor multi-target tracking against a background, which has a high demand of real-time and a relative demand of tracking precision, and analyzed the advantages and disadvantages of the existent classical algorithms theoretically in this environment. Based on the research and the analysis, general association algorithm has been extended into multi-senor system through the parallel structure and a new algorithm named parallel centralized multi-sensor general association algorithm has been proposed. In this algorithm, a lot of hypotheses are built with the measurements of each sensor and the track of each target, the score of every hypothesis could be obtained with the formula of the score function in multi-sensor general association algorithm, and the state estimation of the fusion center is gained finally. The simulation results show that this algorithm could track the targets effectively with the high density clutter and the general performance of this algorithm is better than that of sequential centralized multi-sensor joint probabilistic data association algorithm and parallel centralized multi-sensor probabilistic nearest neighbor standard function.