Research On The Algorithm Of Iris Recognition

With the rapid development of communication technologies in the Internet age, human identity requires more accuracy, reliability and security in modern society, and traditional identification methods cannot meet the need of reality. Therefore, biometric identification technology is emerged. Iris identification has more advantages in stability and uniqueness. Iris recognition technology is concerned by more and more scientists. Now the theoretical research of iris recognition has been continuously improved. At the same time, it has been used in many application areas.In this paper, the background and significance of biometrics are introduced firstly. Several existing biometrics are briefly introduced, and their advantages and disadvantages are compared; then the technical characteristics, development history and application area of the iris recognition and the entire frames of the iris recognition system are introduced. Finally, the algorithm of iris recognition is studied at each stage. On the basis of insight into the common iris recognition algorithms, research and the improved algorithms are as follows:1. A fast iris location algorithm based on least squares fitting is studied. Firstly, the iris image is filtered in the algorithm; and the outside area of pupil is eliminate by the way based on histogram threshold method; then the pupil area is projected on the ordinate and abscissa, and a preliminary diameter and center coordinates of the pupil are identified. Using Canny operator to extract the pupil edge, and then a least squares fitting is used to fit a circular arc. The inner edge of the iris is located. For iris localization of the outer boundary, the outer edge of the visible arc is firstly estimated, and then the interference caused by the eyelids end eyelashes is eliminated, and the least squares is used to fit the both right and left sides of the arc. in this way, the iris outer boundary is located. Finally, the noise statistical method in horizontal regional is used to locate the eyelashes. Experimental results show that the algorithm proposed is more faster and accurate.2. The iris image is normalized by improved line extraction. This method is used a way based on the average point on the lines to normalize the iris area. Iris normalization used the traditional line extraction method may be a mistake if the center of inside and outside circle is not in the same level. On the basis of insight into the basis of the principle of the line extraction method, the coordinates of the point on the circle is represented by the center of iris inner and outer circle respectively. This method can ensure that the iris region in the radial direction of the feature point extraction are corresponded, thus iris area is normalized accuracy.3. An iris feature extraction algorithm which extracts the local texture feature based on the Center-Symmetric Local Binary Pattern (CS-LBP) is studied. The algorithm calculates the difference of gray value for each symmetrical pixels surrounding the center pixel, and the difference is quantified, and then the arranged values in a particular order is as eigenvalues of the center pixel. The resulting eigenvalues range from0to16, so the dimensions and complexity of the calculations are decreased.4. The matching part of the iris is divided into two parts with eigenvalues match and ratio statistics. Firstly, the corresponding eigenvalue of two templates are compared, and0is assigned to the same results and1is assigned to the difference results. Then having a statistics the value of1accounts for entire feature template, and searching the minimum distance of iris codes as the basis for the final match after cyclic shift. The best threshold and cyclic shift digits are determined by the experimentation to improve the match rate.The algorithms designed in this paper are applied to the sub CASIA4.0iris database, and all the above algorithms are simulated on the platform of Matlab11b. Experimental results show the iris recognition algorithm proposed has higher recognition rate and a certain application value. It can be used as a reference of the theory in the field of iris recognition.