Study On Multi-sensor Target Recognition Based On Intelligent Computing

In recent years, along with increasing of sensor category, as well as the continuously raising of sensor function, only depending on one sensor has already can not satisfy the demand that the target recognition. Therefore, the multi-sensor data fusion technology used in target recognition has become a research hotspot. This paper introduced the multi-sensor data fusion in several intelligent computation ways that used for target recognition. The analysis explained that the D-S evidence theory respect combines with BP neural network and fuzzy neural networks that used for target identification the advantage had by the latter.D-S evidence theory is the most commonly method to deal with the uncertain problems in multi-sensor data fusion. But the evidence theory is only applicable to the situation of independent evidence. For the fusion of target recognition based on D-S evidence theory, acquire the basic probability assignment is a subject closely related with application, and its also the difficulties in practical application. This paper is combined together evidence theory and the neural network to handle this problem and make use of neural network to seek basic probability assignment.The neural networks have been widely used in various fields, of which the most widely used is the BP neural network. Because BP neural network has local minima, slow convergence and other problem, this paper put forward the method that combine BP neural network and fuzzy inference which called fuzzy neural network to seek basic probability assignment. Fuzzy neural networks has significantly broadened the scope of neural networks and the ability to process information. At the same time, self-learning of neural network makes the automatic extraction of fuzzy rules and membership functions can be achieved automatically.Using fuzzy neural network can improve the whole system learning and presentation skills, the recognition rate higher than use BP neural network alone. The finally simulation results show that the combination of neural networks and fuzzy reasoning can acquire the better basic probability assignment function, and combined with D-S evidence theory, can get more accurate recognition results, the results from fuzzy neural network in conjunction with the D-S evidence theory in target recognition were excellent to BP neural network combined with the D-S theory, showed that the former has some values.