The Algorithm Researching Of Iris Recognition Technology

With the development of biometrics for human identification, iris recognition has received increasing attention in recent years. Several algorithms were presented and proved to be effective. The whole process of iris recognition is composed of three steps named edge detection, feature extraction and matching, respectively. A new iris recognition algorithm is presented in this thesis.Firstly, a two-step locating algorithm named cursory locating and accurate locating respectively is adopted. The radius and center of gravity of the pupil is found after isolate pupil from iris image using dichotomy processing. To detect the outer boundary of iris, canny operator is adopted in a smaller region to locate iris's center and radius. Compared with other methods, compute consumption is reduced in this method, and thus the detecting speed is improved.Secondly, a method based on Daubechies-4 wavelet transform is discussed to extract iris's feature. After divided iris into ten strips according to its texture distributing, each strip is processed with Daubechies-4 wavelet transform. The feature value is denoted by the mean and standard deviation of every window's coefficients of wavelet transform, and 112 feature values can be ascertained for every iris, whose storage space need only about 1K to 2K.In classification phase, Euclid distance is adopted as the classified tool. And the classified standard is the mean of the Euclid distance vector.Lastly, something must be noticed is that this algorithm is robust over scale, rotation, and move of eyes. In order to evaluate the proposed algorithm, some test was taken on the CASIA database and compared the result with others. The result indicated that the proposed algorithm is effective.