Iris Image Preprocessing And Extraction Of Key Points

Iris recognition is an important component of biometric identification technology. There is very high accuracy, reliability and robustness and more information can be distinguished. Through analysis of the rich iris texture feature, combined with the corresponding feature extraction and classification method can achieve the iris identification. However, due to the imaging process of the iris by the acquisition equipment, light, drift, rotation, scaling and other factors, makes iris recognition as a pattern recognition and machine vision in the most challenging one of the topics.In this paper, by systematic analysis and finishing the work of our predecessors on the iris recognition results obtained on the basis of iris image preprocessing, iris feature extraction and coding, pattern matching and classification of the three thematic design of the algorithm for iris recognition technology to do an in-depth research. In the iris image pre-processing, offering improving fast algorithm based on statistical analysis of the threshold, first a rough estimate of the center of a circle with a radius of the pupil, the establishment of coordinate system, in accordance with the characteristics of the pupil is calculated with the edge of the iris and the center of a circle, and then use the binary iris image and the coordinates of the pupil, using the edge of the search, find the outside edge of the iris. Compared to Daugman algorithm and the enforceability of the algorithm is very good, and significantly reducing the computation to avoid the blindness of the search, the application of strong experimental evidence of the effectiveness of the algorithm. Algorithm for iris images precise positioning, so that after positioning of the iris image with the rotation, translation and scale invariance. To overcome the current prevalence of the problem of excessive algorithm, simulation results show that the iris to improve the image pre-processing to achieve the desired objective. Iris image of the traditional method of normalization has been improved, according to both within and outside the edge of the iris center are inconsistent, so that after the polar coordinates of the iris image with rotation invariance and scaling invariance of the pupil, the elimination of human eye pupil rotation and zoom the image the impact of identification. 2-D filters for the convolution operation slow computing, using two-dimensional fast Fourier operation on completion of the iris feature extraction and encoding, improved speed of operation.