The Research Of Identification Algorithm Based On Iris

The iris recognition technology is considered to be one of the most promising biometric authentication technology because of its high stability and security, and has been widely used in financial,electronic commerce,defense, and other areas of the tip.This article study in improving the performance of the algorithm in the preprocessing, feature extraction and pattern matching of iris recognition technology. The specific research contents of paper have some areas as the following:In phase iris image preprocessing method of the edge is detected via an improved Canny operator is proposed in this paper. The location method of iris’ s inner boundary is that binarization and morphological combination method, then introduce the adaptive Canny operator and a least squares combination method which to extract the outer edge of iris. Normalize the preprocessed iris image to the rectangular area of 512′64 size with the double polar coordinates before texture feature extraction. Use histogram equalization to enhance important texture feature.In phase texture feature extraction method of 2D Log Gabor filter is proposed in this paper, Design a set of 2D Gabor filter groups with four dimensions and four directions imaginary part only in 16 channels aim to extract feature in multiple perspectives. The Log Gabor filter groups which are used to multiple perspectives of iris images, and the introduction of PSO parameters were optimized filter set, after images which filtered are encoded with zero crossing detection. And then in the pattern classify the iris features, using the new method based on SVM and Hamming distance. Finally the MATLAB GUI module is applied to design the iris identification interface. The proposed algorithm is used to carried out experiment with the CASIA database on iris image. Experiment results show that is a lower standard deviation false accept rate and error rejection rate were reduced to 0.01% and 0.32%,the recognition rate of this method reached 99.67% compared to classical methods. That is the proposed algorithm in the paper has better iris recognition rate, which can obtain a better identification effect.