Multi-sensor Target Tracking Data Fusion Theory And Algorithm

As a new and developing crossed subject, study of multi-sensor data fusion aims to how to combine information from multiple sensors and obtain more accurate estimation and inference. Multi-target tracking with multi-sensor data fusion technology is the leading technology in science research now, and the multi-sensor information can be used to increase the precision of target state estimation in this technology. It has more advantages than the case of using single sensor.The basic principle, fusion model and common method in data fusion are introduced, and the Kalman filter and the expend Kalman filter used in target tracking are introduced. As the important problems of multi-sensor data fusion, track association and track fusion are researched deeply based on these theories.Dissertation compares the serial structure and parallel structure of multi-sensor join probability data association (MSJPDA) arithmetic, and points the performance of serial structure is slightly superior to the parallel structure with the number of sensors increased. Classical distribution fusion arithmetic used in distributed association structure has been studied and solved with genetic algorithm for decreasing the computational burden. The local tracks from different sensors are always asynchronous. To solve the problem, this paper presents a multi-sensor asynchronous track association algorithm. The simulation results illustrate that this new algorithm can solve the problem of asynchronous track association effectively, and the association correct rate approaches 90%.Dissertation researches parallel and sequential arithmetic used in centralized fusion structure, and derives out optimum algorithm used in the distributed fusion structure based on this. The algorithm has optimum performance in cost of aggravated calculation burden. The simple fusion method (SF method), Bar Shalom-Campo fusion method and adaptive fusion method have been studied. For the adaptive method can adaptively select the appropriate method according to the systematic demand, computational burden of adaptive method is lower than Bar Shalom-Campo method. The adaptive method approaches the Bar Shalom-Campo method in the fusion precision. The simulation results validate the conclusions.