Research Of Track Fusion Algorithm Of Distributed Multisensor System

In the multi-sensor information fusion system, centralized and distributed fusion structure is frequently been used, which, comparing the distributed fusion structure and centralized fusion structure, it except has a similar fusion in addition to precision, also has a lower construction and maintenance cost, higher reliability and availability, even can reduce the data bus bandwidth and lower data processing requirements and so on. Currently, when designing multi-sensor multi-target tracking system, the distributed fusion structure has become priority application. And track fusion is the final part of the multi-sensor information fusion technology, it is also the important and difficult part in such research area.Aiming at the present state that several mainly used track-to-track fusion algorithms of distributed multisensor tracking systems are inevitably needed to calculate the cross-covariance of error, it puts out a algorithm which uses ELMAN ANN to compensate fusion error caused by cross-covariance after fusing partial tracks by covariance convex (cc) algorithm. Experiment results illustrates new algorithm can efficiently solve the fusion problems caused by common possess noise which is under multisensor environment.