Study On The Track Correlation Algorithm In Distributed Multisensor Data Fusion System

Track correlation algorithm is the key technology of data fusion system. The classicNearest Neighbors Algorithm (NNA) is discussed, and because of its shortcomings, anew Global Best Track Correlation Algorithm (GBTCA) is proposed in this thesis. Inorder to evaluate the performance of the new algorithm, a complete data fusion systemis also implemented. The main work is as follows:Because of the highly dependent on threshold and leak of consideration of globalsolutions, the classic Nearest Neighbors Algorithm will have some issues when thedensity of targets is high. To solve this problem, global search strategy, gating anddynamic programming are used to build a new correlation algorithm-the Global BestTrack Correlation Algorithm. Two experiments are given. Both of the two algorithmsare ran, and compared to NNA, the result shows that GBTCA has a higher correctcorrelation rate and less dependent on threshold values.This thesis also gives the details of a data fusion system design, from the targetstoring and updating strategy to the methods of space and time alignment, involving allaspects of the system design. It is very valuable for the development of similar system.A real data fusion system is implemented, which is used to evaluate the performance ofthe new algorithm.