Multimode Biometric Authentication Of The Prototype System Research And Design

Biometrics are popular technologys in the most recent decades. It can be used in access control, security, identity authentication, file management, intelligent human-computer interaction, etc., and has a very wide range of applications. And it include the biometrics collection, graphic image processing, signal processing, pattern recognition and so on, so there are a lot of research values. Biometric technologies are mature, but they are all based on single-mode characteristics. Biometric technologies based on single-mode characteristics have less information entropy than multi-mode biometric identification technology, so the correct rate of Biometric technologies based on single-mode characteristics would be lower than multi-mode biometrics identification technology. Therefore, the research based on multi-mode biometric features has mroe significanance.Based on the traditional recognition technologies, to the information of the human face and iris information, this thesis'work are:1. To some problems in the traditional iris recognition technologies, this thesis presents iris Image Segmentation, a method of spot filling and round position. The round position is improved from gray projection. These methods are all proved by experiments.2. To service for an authentication Model, this thesis improved the traditional Gabor wavelet filter.3. In the part of facial feature extraction, there is a problem that the datas are large. This thesis presents a model that feature extraction processes twice. Firstly, the improved Gabor wavelet filter is used to extract the underlying characteristics, sceondly, KICA is used to extract High-level feature. This method make the feature datas smaller and are proved by experiments.4. In this thesis, PDBNN is researched, and experiments prove the advantage.5. In this thesis, all the framework and modules in this system are researched.