Research Of Iris Recognition Based On Wavelet Analysis

With the development of information technology in contemporary society, the demands for information security are becoming more and more important. As the technology of biometrics identification has great advantages, it has been improved and applied in many fields. Iris has some special advantages such as uniqueness, stability and no-infringing comparing with other numerous biological characteristics. Iris identification technology is regarded as one of the most promising biometric cognition technology.An iris recognition system based on iris patterns is composed of image pre-processing, feature extraction and coding, matching and recognition. The iris images of pre-processing, mainly includes three parts: locating in-and-outer boundaries of iris images, normalization and image enhancement.At present there are some mainly methods, such as Daughman's iris identification using differential operator locating inside and outside edge of iris, 2-D Gabor filter for encoding iris texture feature; Wilds's multi-scale algorithm in matching. Boles proposed iris identification algorithm which is based on the zero-crossing detection of wavelet transform.On the basis of summarizing the main technical research of the iris identification, iris boundary localization method was discussed in the paper. For the problem of slow speed in traditional locating algorithm, this paper proposes a new iris location algorithm which is based on the human eye geometric structure. The locating process which contains computing average grayscale and searching boundary points by detection template combines the geometrical characteristics of the eye image. The presented iris locating algorithm is improved both in speed accuracy, and performs with good recognition rate on the images CASIA iris database. Compared with two classical methods, the proposed method ensures the probability of successful localization and reduces the time consumedly. The correct locating rate of this method is 97.63%, higher 4.915s than Daughman's location, 0.559s than Wilds'Hough transformation. In feature extraction, a method based on Sym6 wavelet transform and singular value decomposition is used to extract iris's feature. In pattern match and recognition, weighted Euclidean distance criterion and the average minimum distance method is used in feature matching of iris. Each image is chosen as testing image and clustering feature vectors in the CASIA iris database which is provided for matching by criterion mentioned above. The correct recognition rate of this method is 97.31%.This method is proven to suit in practice by the experiment results. The results of testing experiments indicate that the algorithm is reliable, efficient.