Research On Target Tracking Algorithm For Multi-sensor Information Fusion

Multi-sensor information fusion is an emerging and crossed discipline, multi-sourcesinformation is integrated to achieve better performance than any single sensor. Thetechnology in the military and many civilian areas have a wide range of application prospects.Multi-sensor target tracking is a paradigm of information fusion technique in the field oftarget tracking technology, and has become one of the hot current scientific researches.Firstly, the thesis expatiates on the significance, research status home and abroad ofinformation fusion research. Then it discusses the function model and the structural modeland so on in the field of information fusion. And it also carries on some key technologies,such as the basic principle of multi-sensor target tracking, target maneuvering model setup,tracking gate formation, data association, tracking initiation and termination.The paper focuses on the optimal and sub-optimal filtering algorithms for handling avariety of linear and non-linear models, including the Kalman filter, extended Kalman filter,unscented Kalman filter and particle filter. At present, most of the studies were carried on asingle target tracking filter estimates. This paper not only use different filtering algorithmsthat can filter and estimate the linear and non-linear and non-Gaussian state of single targettracking, but also can using these different filtering algorithms for multiple targets, whosestate are (strong)maneuvering and trajectories crossing. Then, maneuvering target trackingfiltering algorithms are simulated by Monte Carlo, and the simulation results indicate thesealgorithms are efficient and they can improve the accuracy of target tracking. At the sametime, we can analysis、compare the advantages and disadvantages and application scenarios.The complexity of the target movement form and environmental interference makes thetarget model showing diversity, in order to solve this problem, IMM(Interacting MultipleModel)algorithm is researched and analyzed, and combined with probabilistic dataassociation algorithm, the formation of a multi-sensor-based interactive multi-model fusiontarget tracking algorithm is completed. The simulation result tests and verifies that thealgorithm has better tracking effect for strong maneuvering targets.At last, the thesis summarizes all research work discussed above, and point out the futureresearch direction in this field.