Fingerprint Recognition System Based On Neural Network

With the developments of the information age, biometric identification technology has become a hot identification technology, automated fingerprint identification technology as a very important one branch has been widely used in public security, customs, banking, Network security needs identification field. However, as the fingerprint image noise and skin elasticity, and other factors, and the need for a large number of fingerprint matching, Automated Fingerprint Identification System identification rate is not high computational speed slow, and so on. This paper attempts to neural networks and pattern recognition technology, a number of effective new methods, To solve the Automatic Fingerprint Identification Technology of the difficulty of finding practical ways. To sum up, this paper fingerprint identification system to the following aspects of in-depth research.The technology of fingerprint image preprocessing, the image of the original fingerprint image enhancement, using filtering direction of a fingerprint background separation; Local Gray variance integrated fingerprint image segmentation algorithm and the image of a fingerprint segmentation algorithm orientation information. A new complex has been modified and improved, effective, the fingerprint image segmentation; Refinement of the image of a fingerprint technology, in comparison, optimization, The algorithm used for the rehabilitation of the ridge and Data Processing. After pretreating, the fingerprint image quality has been markedly improved, to better meet the needs of the follow-up work.In feature extraction technology research, the paper extract the fingerprint classification and identification of characteristics. Using innovative global features for the neural network fingerprint classification, and the extraction of partial fingerprint minutiae as the characteristics of neural network fingerprint identification features thus effectively improving the overall performance of the purpose of identification.In the fingerprint classification study, this paper, a fingerprint recognition algorithm based on BP neural network. Neural network models were used at different stages of fingerprint classification and identification. Experimental results show that the method of obtaining a better recognition results. Finally a brief introduction of BP neural network-based CPLD hardware. In this paper, in-depth study of the automatic fingerprint identification image preprocessing, feature extraction and fingerprint identification methods and the theory, focus on the fingerprint image binarization processing, feature extraction and fingerprint classification algorithm, for the further development of commercially valuable Automated Fingerprint Identification System to provide a theoretical basis and technological foundation.