Research On Iris Recognition Preprocessing

With the rapid development of the information and communication technology and the increasing popularity of network technology, information security has become a hot topic nowadays. And people pay more and more attention to identity recognition technology. As the traditional methods for personal identification have been unable to meet people's needs, a new method, biometrics technology, is perceived by everyone. Because of the property of stability, reliability and uniqueness, biometric technology is becoming an active topic. Iris recognition is a new biometrics technology and is considered to be the most reliable one. It has wide development prospects and great application values; therefore the research of iris recognition has very important practical significances.Mainly, iris recognition system includes the following parts:iris image capturing, image preprocessing, feature extraction and matching. Image capturing is to get original eye image using capturing device. Image preprocessing is to process the eye image so that the image can be used for feature extraction and matching. Feature extraction is to extract features benefited for recognition from the images after preprocessed. Matching is comparing and classifying the features with some similarity measures. Image preprocessing is one of the most important parts, and the result of it will directly affect the accuracy of iris recognition. Image preprocessing includes localization, eyelid and eyelash detection, normalization, etc. We focus on eyelid and eyelash detection in the preprocessing process, with which the localization process is guided.This thesis modifies Wildes'eyelid detection algorithm on the basis of the summary of the existed methods. The edge detection and Hough transform of parabola algorithms are used for eyelid detection and localization in the Wildes' recognition system. But for some images, fitting with parabola may lead to great errors. In this thesis, Canny operator is used to detect the edge information instead of gradient operator; upper eyelid and lower eyelid are detected using Hough transform by being divided into two segments, respectively. The experimental results show that the modified method can detect eyelids more effectively.This thesis proposed a new eyelash detection algorithm based on Max filter. Max filter is one type of the order-statistics filters. It orders the pixels in the filter window by the gray value and replaces the gray value of the filtering point with the max one. In the difference image between the original and the filtered image, the point which has a small gray value in original image will be more obvious. What's more, as the filtering is within the local window, it can remove the impact of the uneven light condition in the image. According to these characteristics, we proposed an eyelash detection algorithm based on Max filter in this thesis. Firstly, we filter the image and get the difference image;then, we detect eyelash pixels from the difference image using dynamic threshold and connective criterion. The dynamic threshold of the detecting point is decided according to the gray values in its neighborhood. And the connective criterion means that every eyelash point should be connected with another eyelash or eyelid point near it.Meanwhile, iris localization process is modified by the proposed eyelid and eyelash detection methods. With the location information of the eyelid and eyelash detected, the disturbance information is removed and the iris can be localized more accurately. Also, the preprocessing process is redesigned more reasonably.