Modular Neural Network Research On Fingerprint Recognition

With the development of the computer.identity identificiton technology by computer make great progress.As a branch of intelligence technology, the development of neural network can promote the development of pattern recognition.But the result of application is always due to its operators because of the lack of strict theoretical system. Although we have known that one single hidden front feedback network rest with approaching any complex functions, we still cannot find ideal corresponding network structure in practice so that it influences the general capability of network. Nowadays, the experts of neural network have used many ways of mutual cooperation and competition to construct the studying system of neural network. From a large number of theories and experimental studies, multi-neural network is better than single-neural network in stability of general capability; it has great visual ability and it is unlike the former disposal, in order to get better result the structure of network is always determined by operators' experience or a large number of experiments.In all, this article has made thorough study in the optimal combination of modular neural networks, predisposal of fingerprints, extraction of features and identification of characters and fingerprints based on neural network.In the study of fingerprint, on the basis of analyzing and studying modular neural networks, with the known methods based on details of fingerprints, the method of neural network has been reasonably used so that it effectively overcome the effect of fingerprint deformities, increase the resistibility to noise and deduce the reliance of fingerprints location.