A Multiplex Biometric Verification System Based On Fingerprint And Face

It is personal identity authentication that biometric identification technology is, which combining with computing and all kinds of optical principles, using high-tech biological means to obtain people's physiological or behavioral characteristics. Because do not need to remember complex passwords, nor need to carry other things around them except themselves, biometric identification technique has become the most promising candidate. Biometric identification technology is not only fast and reliable, but also can integrate with the security, monitoring and management system easily to achieve automatic control by means of computer. Therefore the technology has broad application prospects and gained widespread attention.In this thesis, vast literatures about the hot spot of biometric identification technology were reviewed. A variety of single biometric technology was compared. And the existing principles of biometric technology, algorithms and application prospects were discussed. However, due to its disadvantages and limitations, the single biometric technology can not satisfy the increasing security requirements. In order to improve the security and applicability of the biometric system, multi-biometric identification fusion system is proposed. In this context, a multiplex biometric verification system based on fingerprint and face was put forward and investigated in this thesis.Firstly various algorithms of feature extraction and matching, pretreatment techniques of fingerprint or face recognition system were studied respectively. Then in decision-making level, the algorithm between fingerprint and face was fused using a linear method referring to the matching scores of a single biometric feature. Fingerprint verification subsystem, face verification subsystem and multi-biometric identification fusion system of fingerprint and face was built up respectively. The main contributions are as follows.1. The algorithms of two single-mode authentication technology, based on face, fingerprint identification technology, are studied and improved to increase the recognition rate of single-mode authentication system.In fingerprint verification subsystem, the algorithms of normalization and calculating orientation field are introduced first, and then the improved OPTA (One-Pass Thinning Algorithm) is applied to thinning binary image of the fingerprint image. Next, we use Poincareindex method to extract singular point of the fingerprint image. Finally, feature images matching is conducted on Hough transform. The results of the fingerprint verification subsystem are satisfactory.A method of face feature extraction based on 2D Gabor filter combines with the three-dimensional face matching algorithm to form the face verification subsystem. The built algorithm is applied to standard face database in MATLAB language environment. The experiment results of the algorithm show its superiority.2. As for different mechanism of multi-mode biometrics fusion, in this thesis, the linear fusion approach is applied in decision-making level between two biometric verification systems. The method that we give different weights in decision-making level according to matching score of the single biometric feature combines the advantages of fingerprint recognition technique and face recognition technique. Therefore, optimal fusion result is obtained.3. The total fusion system combined fingerprint identification subsystem with face identification subsystem is designed and tested by the corresponding fingerprint database and face database. The results show that the above algorithm has better recognition rate and optimize the whole performance of the fusion system...