The Research Of Fingerprint Identification Method Based On The Continuous Classification

The technology of fingerprint identification is a kind of important biometrics style. On the network environment, more and more people participate in e-commerce, e-government, finance and other online working ways. The traditional identity authentication ways, such as keys and password, already cannot satisfy the information security requirements of individual privacy or even state secrets not to be leaked. Each of kind of biological characteristics has its unique advantages, such as stability, permanent, uniqueness and safety, so it can be used effectively to avoid loss, forgotten, reproduction and attacked etc hidden safe troubles. Fingerprint identification, because of friendly contact interface, low price and other features easy to use, is the biggest concern of biometric identification.This thesis summarizes the development status of fingerprint identification technology, analyzes the hot research problems of fingerprint identification, at last, confirms to research the whole realization process of fingerprint identification around continuous classification, and researches how to improve the comprehensive fingerprint matching performance with a consideration of automatic fingerprint identification method as a whole. The fingerprint classification information used in continuous classification can also be used into the matching stage, meantime, the usage of minutiae neighboring structure not only convenient to extract classification sampling points, also conduce to the realization of fast matching. Thesis's innovations are as follow.(1) This thesis researches the whole fingerprint identification system using continuous classification, the aim of doing so is to search every stage reasonable method for the system. The first step is to choose appropriate pretreatment and enhancement algorithm for fingerprint identification system, the selected pretreatment method can satisfy the feature extraction requirements for fingerprint continuous classification and matching.(2) This thesis studies the fingerprint classification method in detail, improve the method of positioning fingerprint reference point. At last, the experiment is designed to inspect the feasibility of the classification feature vector and the performance of the classification method, and the image preprocessed effect also can be tested.(3) The final matching algorithm adopts the fingerprint information getting from classification. Choosing fingerprint singularity gained in classification as the matching reference point can improve the efficiency of positioning the reference point. In addition, the minutiae structure is convenient for setting conditions to reduce the minutiae matching computation of the template, and also can reduce the fingerprint matching time, make the matching speed get significantly improved finally.The research results show that the choice of pretreatment method for continuous classification can satisfy the demand of classification feature extraction, using the Gabor filtering as the image enhancement algorithm is to consider fully the request of the classification feature, the algorithm not only can realize image enhancement but also facilitate ridge line width calculation. The extracted classification feature is so effective that it can set down the candidate fingerprint amount to the acceptable range. The improvements to matching stage can reduce the number of two minutiae pairs to match and the introduction of new properties can eliminate two apparent dissimilar minutiae to further match. In a word, this method eventually lower the computation cost so as to reach the purpose of improving the matching speed.