|
Modern society, with the rapid development of information technology, the amount of data increases exponentially, more and more new technology into people's daily life, more and more people are able to feel the digital life has brought us warm and convenient. Early on, in order to identify a person's identity, the traditional approach is to use a password or card for authentication, this traditional technique to a certain extent, solve the basic problem of identification. Since traditional identification technology inherent defect, people began to turn to research-based biometric identification technology, compared with traditional technologies, emerging technologies that can really make up for its flaws and shortcomings. But also has good biometric uniqueness, stability and universality. In recent years, more and more is used as a biometric identification technology, in addition to the early fingerprint, face, iris, as well as hand-type, palm prints, signatures, etc. voiceprint. So many biometric can be used, but also makes many researchers turned their attention to the multi-modal identification technology research in the field. In this paper, face recognition and fingerprint recognition technology, in-depth study of the multi-modal feature fusion theory and methods are given based on two feature fusion biometric system design. The main research work and achievements are as follows: (1) in face detection research, this paper based on adaboost human face detection technology has been improved to achieve by positioning the human eye and face position, accurate detection the face region and the face pose algorithms. (2) for the face recognition technology for the in-depth study, to achieve a feature based on PCA, LDA feature, GABOR features, LBP features face recognition algorithm, algorithm parameters for each of the repeated experiments and adjusted to achieve a better results. (3) of the fingerprint recognition technology conducted in-depth research, detailed discussion of the direction of filtering, image segmentation binarization, thinning algorithm improved fast fingerprint image preprocessing algorithm. Preservation of the minutiae extraction method and removal of the dummy feature points. Feature matching method and a systematic study, including the fingerprint minutiae data structure, the initial realization of the matching, and the secondary matching the coordinate adjustment, global matching methods and matching judgment conditions. (4) on the basic theory of information fusion in-depth discussions, and LDA face features, GABOR characteristics, and the geometric characteristics of the fingerprint modal characteristics of these three methods for fusion research on the human face features and LDA GABOR feature fusion algorithm is studied, based on the realization GCCA feature level fusion algorithms. Then again after fusion of facial feature characteristic fusion algorithm and fingerprint conducted in-depth studies were implemented using the adaptive weighted and DS evidence theory the match-level fusion algorithm. (5) proposed a multi-modal identification system design, describes the structure and parameters related to the system design, including databases, files, structural design, and three specific functional block diagram of the module. Also proposed an extended class design methods to improve the robustness and stability of the system platform.
|