An Improved Recognition System Based On Multi-characteristic Of Finger

In recent years, Information Security has been a hot topic and gains more and more attention. Authentication is one of the most important part of information security, and the most efficient method for personal identification is based on biological features. The reason why such method is efficient is the biological features have characters of individual uniqueness, and they are difficult to copy and steal. Biometric recognition system has multiple application on different field, such as fingerprint and iris has been used in access control system, face and gait has been used in public security field, and DNA has been used to find crimina, etc. When the biological features are used in real situations, the constrain of each features should be concerned, which indeed limit the application of the single biometric recognition technology. To solve such problem, the multi-biometrics/multi-modal recognition system is proposed.Multi-biometrics / multi-modal recognition technology is combine two(or more) kinds of biological features to achieve the final result. The fusion process can happen in the any stage of the recognition system, which determine the type of fusion. The different type of fusion is described as bellow: sensor-level fusion, feature-level fusion, score-level fusion and decision-level fusion. The multi-biometics recognition technique have two strong points, one is it cover the shortage of single biological feature’s limitation is real applications, the other is such technology is implement by the improve of algorithms instead of increase the precision of device, which make the thchnology is valid and feasible.This article achieves a finger-based multi-biometrics recognition system. In consideration of the convenience and comfort in picture collect progress, this system adopt fingerprint、finger-knuckle、finger-vein three biological features, and use the strategy of score-level fusion to achieve the final recognition system. Our work is done in MATLAB platform. Until now, we have achieving the following main result: 1) Layer seperation algorithm is added to the preprocessing period of finger-vein recognition, which lead to an enhanced finger-vein image and a new kind of soft feature; 2) give the feature extraction method of the new soft feature; 3) in fusion stage, a new weighted fusion algorithm is proposed, in which the weights of finger-vein is given according to the mean value of the new soft feature. Experiments show the whole multi-biometric recognition system work efficient, and the final result is good.