Abstract: In a multi-node distributed multisensor fusion system, the problem of track correlation can be transformed to a problem of multi-dimension assignment. The problem of multi-dimension assignment is a typical combined optimization problem, it is very hard to obtain the optimum solution, and its computing burden is easy to increase exponentially with the increase of the numbers of targets and dimensions. A three-dimension mean-field neural network model is proposed to solve the problem based on the two-dimension mean-field neural network. The experimental results illustrate that this network model can solve three-dimension assignment problem effectively, and that the correct percent of track association is high. It can satisfy the practical needs when the number of target is not very large. The three-dimension network model proposed in this paper can be generalized to the condition of multi-dimension.

Key words: track correlation, multi-dimension assignment, mean-field neural network, information fusion