Fuzzy Clustering RBF Neural Network Applied to Signal Processing of the Imaging Detection

Imaging fuze can get the geometric shape and exterior features of target in very close distance, and supply the information for target recognition, so it is help to improve the destroying efficiency to enemy targets. But because of the large amount of the information and the real-time requirement, the image of target is always distorted and incomplete. In order to recognize the encountering conditions in imaging detecting, a method based on fuzzy clustering radial basis function (RBF) neural networks is presented. The fuzzy clustering RBF neural network has more adaptive to recognize the encountering conditions. A fuzzy C-means clustering method based on minimizing the mean square error in one category is adopted to determine the RBF layer, and the grads of membership degree is used to determine their shape factors. The theoretic analysis and experimental results show that, the fuzzy clustering RBF network has more powerful generalization ability than the normal RBF network.