Iris And Face Based Multi-Biometrics Fusion Identification

There is no perfect biometric system on the real world. Each has its own strengths, weaknesses and vulnerabilities. However using fused biometrics on a higher data capacity, counterfeit-resistant, secure identification card will lead to the creation and implementation of the trusted identity card required for our national security. At present, it is well known that the iris identification is the one of the best accuracy in biometrics, as well as its high rate of failure to capture and failure to enrollment is the big challenge to large scale practical real applications. Meanwhile, there is easy to capture face image friendly to the face identification, though this biometric performance is not so satisfied. It is an interesting combination point.In the dissertation, we discuss multi-biometrics fusion identification based on iris and face from the capture layer, the feature extraction layer and the decision layer. Especially in these aspects as follows: the real-time fusion pre-estimation of iris images, the two dimensional projection based iris localization, the face template fast match by the mesh degree, the subjection function based pagoda layering fusion method and the iris feature template data hidden method based on the bits-flowing. The goal of the selection with these two kinds of intactness capture biometrics is that using face image abundant information maintains the global accuracy same as the iris identification under the condition of iris images quality descending for debasing the iris's failure to capture rate and failure to enrollment rate. Moreover, we do some researches in the security of biometrics system self and propose a method to promise iris template data safety that iris template data are encoded in face template data bit-by-bit to create a fusion biometric template for store and transmission.Finally, we develop a practical multi-biometrics system based on iris and face to verify these methods and to investigate the real application commerce potential of this system. The experimental result is shown that the system is valuable to accelerate Biometrics into real security applications.