The Research On Several Key Technologies Of Iris Recognition

With the rapid development of information technology, the research on informat- ion security has become one of important topics. The identity recognition based on the biology characteristics of human body becomes more important. The iris recognition is a kind of the biometrics technologies based on the physiological characteristics of human body, compared with the feature recognition based on the fingerprint, palm-print, face and sound etc, the iris has some advantages such as uniqueness, stability, high recognition rate, and non-infringing etc. Hence the iris recognition technology has recently become popular in identity recognition.In this thesis the developing situation and applying prospect, the advantage and disadvantage, and the basic principle about the iris recognition technology were introduced. By analysing and comparing the several iris recognition methods, the research on several key technologies in allusion to the iris localization, the normal- ization, the feature extraction, the pattern matching etc was carried on, the main work was accomplished in the paper as following:Firstly, a new iris localization algorithm was proposed. The Canny operator of edge detection was used to carry on the cursory localization to the inner edge of iris image; then, the Hough transform was made use of detecting the outer edge of iris image; finally, the Daugman operator was adopted to carry on the fine localization to the iris image.Secondly,by the wavelet transform being used to extract the texture characteristic of the normalized iris image, the two methods of feature extraction based on the Haar wavelet and the wavelet packet were suggested in this paper.Thirdly,support vector machine (SVM) was used to carry on the pattern matching about the code of the iris feature, because SVM based on the concept of structural risk minimization performs pattern recognition with good performance for two-class probl- em.The experimental results indicated that the iris recognition methods proposed in this article were easily realized, and had good efficiency; moreover, it could overcome these problems such as revolving, drifting and scale etc existed in the present iris recognition algorithms.