Iris segmentation and recognition for human biometric systems

A common way for person authentication in biometric recognition is iris identification, one of the utmost stable and suitable biometric technologies. Nowadays there is a big interest on using iris recognition in visible or in near-infrared spectrum, on smart phones and tablets, to authenticate a person. Generally, this process will takes two steps. In the first step, one or even several images of the person's iris are acquired, using digital cameras in visible or in near-infrared spectrum. In the second step, comparison of captured iris images with images stored in an iris database will take place. So accuracy of an effective iris recognizer depends upon how accurately iris region is localized in the images so that eyelashes, eyelids will be accurately remove from the iris region as they effect the features of the iris. In this work, an effective iris localizer has been proposed which uses number of process such as contrast enhancement, edge enhancement, segmentation, clustering, eyelid detection etc. In this, Snake contouring is used for detecting the boundary of the iris which has been assisted by CHT algorithm to detect the rough iris circle. After that sclera region has been detected and both eyelids has been detected. Finally upper and lower cut-off has been used to exclude the eyelid eyelashes in the detected iris. The proposed algorithm has been tested on UBIRIS database.