The Research And Implementation Of Iris Recognition Algorithm

With the development of information technology and the increasing demand for security, biometrics recognition based on iris, a high reliability and non-offensive bio-information recognition technology, is becoming an applicable method. Iris is the only one in the foreseeable external human internal organs. The surface of iris is rugged, and is composed of many recess, folding and pigment spots, containing rich texture information. These complex structure and texture characteristics makes iris unique, as a highly reliable media for identification, and thus makes iris popular in the researching field of biological identification.A system of iris recognition mainly includes four parts:iris image acquisition, image preprocessing pretreatment, feature extraction, match and recognition. It involves many disciplines and domains, such as computer vision, digital image processing, wavelet theory, pattern recognition, etc. Iris location and feature extraction are key parts of an iris recognition system.In this paper, some key issues related to iris recognition systems are investigated under the consideration of iris physiological structures. The main work of this paper is as follows:(1) The realization of iris image pretreatment. Firstly, a two-step locating algorithm, named cursory locating and accurate locating, is respectively adopted in the process of locating pupil edges. This paper proposes a new method that using wavelet transform and morphology to achieve the approximate position of a pupil. Then the circular integrodifferential operators is used for realizing accurate pupil localization. To detect the outer boundary of an iris, the circular integrodifferential operators are adopted in a given region to locate the iris's center and radius according to prior knowledge. Compared with other methods, computation consumption is reduced in this method, and edge location precision is improved.(2) The realization of feature extraction and pattern matching. Texture feature is extracted by Log_Gabor wavelet. The encoding is realized, based on iris's phase information, while pattern matching is realized, based on Hamming distance. Then an algorithm for realizing iris recognition is proposed, based on region search. This method doesn't need locate outer edges, so it reduces effects on later recognition rates, resulting from faults in locating the outer edges.(3) For the sake of shortening the classification time of iris recognition method base on region searching, an algorithm called "iris index sub_zone" is introduced on the basis of Hamming distance classifiers. In this quadratic classification, if some of the critic features have not been matched, there will be no need to match all the other iris features. By this way, the work load for identifying has been reduced, classification time of classifier has been shortened, and the system identifying speed has been improved to some extend.Finally, the results and conclusions of this paper are given, as well as the comparation between the existed methods and the method proposed in this paper. The experiment result indicates that the algorighm in this paper makes an iris recognition system more robustness and accurate.