| With the increasing people's safety awareness in the 21st century information age, information security has become an important research topic in today's society. Biometric identification technology is one of the researches in information security. As one of biometric identification technologies, the iris recognition is better than other biometric technologies and received great attention and development because of its excellent features.In this thesis, physiological basis, technical characteristics and international developing situation about iris recognition are introduced simply firstly. And then the structure and composition of iris recognition system are elaborated. At last, iris location, image normalization and enhancement, iris feature coding and feature matching are researched individually. In iris location, several location algorithms used widely are described and analyzed, and then an iris location algorithm based on searching small-scale edge regions is proposed. The experimental results show that the proposed iris location algorithm can locate the iris quickly and accurately, and it achieves good location for the iris images of different quality. In iris image normalization and enhancement, Daugman normalization model and transform are described, and a normalized method based on combining Daugman normalization model with transformation of non-concentric is proposed. Then histogram equalization is used to enhance the normalized image. In iris feature coding and matching, several feature coding and matching algorithms used widely are described and analyzed, and then 2D Gabor filter is used to encode the normalized image. Finally hamming distance is used to match the iris feature.Each step in iris location and recognition is studied systematically in this thesis. The results after simulation and validation show that the algorithm studied in this thesis achieves good stability and accuracy and can be the reference for the algorithmic basis of iris recognition system. |
PDF Full Text Request