Research Of Radiant Object Recognition Techniques In Multi-source Data Fusion System

Recently all kinds of sensors is equipped in troops more and more, in which the study of the data fusion of multi sensors makes a rapid progress.The recognition of the radiant point and the platform is dealt with in this paper. By understanding the feature of radar radiant point, one model is provided, which combines artificial neural network, fuzzy matching and DS evidence theory, and which finds the type of radar in three steps.Firstly, one subclass is found by neural network, which is similar with unknown radar. Secondly, the final type of radar is found from above subclass by fuzzy matching. Emphasis is put on the study of the algorithm of Radial Basis Function Network, and of the algorithm of fuzzy matching such as trapezium curve,Cauchy curve,Gauss curve and so on. As a result, the parameter value in the formula is determined. Finally, in order to increase the recognition effect, the information from two or more different time is combined by DS evidence theory. The method of probability evaluation in D-S evidence theory is given.After the type of radiant point is got, using the rules established in data repository to accomplish the platform recognition. The platform can memory the history of the radiant point, and update the information of battlefield environment run-timelyFinally, the system model and test report of system are given in detail. Results show that this model is practical and reliable and can also enhance the recognition radio remarkably.