Based On Multimode Biometric Fusion Identification Method

Identification, which is widely used in bank, public security, and judiciary and all other fields, is unavoidable in our daily life. The traditional single-mode identification system has so many limitations (such as: sensitivity to light, etc.) taht result in practical application of this technology in the many difficulties. With the emergence of data fusion technology, the multi-mode biometrics identification can make use of the complementarities information between different biological characteristics, and integration of the key message of various biological characteristics. It can achieve the goal to synthetically determine personal identity. Many scholars firmly believe that the multi-mode biometrics identification is leading in the future direction of biometrics.The multi-mode biological features data fusion methods, the definition, basic principles, hierarchy and structure are introduced on the basis of common biological features. The fusion and identification of multi-mode biometric are focused to study, and the key technologies of face, fingerprint and iris identification technology are involved. The main contributions are as follows:1) Three single-mode authentication technology, based on face, fingerprint and iris identification technology, are studied and improved in this thesis to increase the recognition rate of single-mode authentication issues. However, there always exist the limitations of single-mode biometric, such as face's sensitivity to light and inaccuracy of iris localization, resulting in low reliability of identification.2) As far as various algorithms of the multi-biometric identification fusion are concerned, the thesis studies fusion algorithm using a Weight-sum in decision-making level. The method overcomes a single biometrics'disadvantages, for example, at collection one of a person's characteristics makes noise impact of the identification reliability. In the database which we use to make tests, the experiment results of the algorithm show a better advantage.3) A new similarity fusion algorithm, based on similarity coefficient model, is studied in the thesis to solve the integration problems of multi-mode biometric features. In the method, the data were fused in the feature layers after extraction and in the decision-making level after pattern matching. As for the test results in our database, the algorithm's effectiveness has been well verified.4) Finally, a multi-mode biometric identification system, on the basis of the new algorithm method above, is built in our thesis. The present research provides certain reference value significance not only for single-mode identification but also for design and development of identification system in practical application.