|
With the extensive application of the eighties of the last century, the rapid development of computer technology and information technology, pattern recognition technology has made considerable progress. Fingerprint its invariance and uniqueness become the most widely used authentication and identification technologies, which involves pattern recognition, image processing, digital signal processing, artificial intelligence, computer, mathematics and other fields, is a comprehensive discipline, widely used in the field of criminal identification, network security, device security, and has important theoretical significance and practical value. The fingerprint recognition technology is not very mature, the main reason is in the process of fingerprint identification, fingerprint image, there are a variety of noise factors, the need for more effective fingerprint image compression, storage, identification and treatment, so the recognition efficiency is not very ideal, the operation speed is very slow. This paper attempts to wavelet technology, neural network technology and pattern recognition technology, made a number of new and effective method to find a practical way to solve the difficulties existing in the fingerprint recognition system. In this paper, based on wavelet technology fingerprint image preprocessing, feature extraction and fingerprint identification method based on neural network for the in-depth study. First, in the pre-processing of the fingerprint image enhancement, segmentation, binarization, thinning and other aspects were carried out in-depth study and discussion of the algorithm, and the simulation results of existing fingerprint image pretreatment, good experimental results prove the superiority of the wavelet technology. In feature extraction technology research, this paper describes a method is based on the template feature extraction method, this method is relatively technology is relatively mature, relatively new feature extraction method based on neural network, but is not very in-depth study pending further study. Fingerprint image recognition, based on the use of artificial neural network model, first introduced the advantages and disadvantages of several traditional identification methods, followed by combination of MATLAB simulation platform for effective training of the neural network, and finally through neural network to achieve the identification of the fingerprint image. According to the analysis of test results of the system performance indicators, the data show that the method has achieved better recognition results to promote the field of fingerprint identification in the future.
|