ANN-based Track Correlation Algorithm in Multisensor System

An algorithm of track-to-track correlation based on neural network is presented . On the basis of membership function in fuzzy logic, this paper presents the method of multi-dimension assignment of track correlation on the condition of multi-node. The problem of multi-dimension assignment is a nondeterministic polynomial complete problem, it is very hard to find its optimum solution and its computing burden is easy to increase exponentially. So a new model of three-dimension neural network is proposed to deal with the problem of three-dimension assignment. The experimental results illustrate that this model not only makes the correct track association percent high, but also the computing time can not increase exponentially with dimensions. And the model of three-dimension neural network proposed in this paper can be generalized to multi-dimension model to solve the problem of multi-dimension assignment.