An Improved Iris Segmentation Technique Using Circular Hough Transform

It is quite easy to spoof an automated iris recognition system using fake iris such as paper print and artificial lens. False Rejection Rate (FRR) and False Acceptance Rate (FAR) of a specific approach can be as a result of noise introduced in the segmentation process. Special attention has not been paid to a modified system in which a more accurate segmentation process is applied to an already existing efficient algorithm thereby increasing the overall reliability and accuracy of iris recognition. In this work an improvement of the already existing wavelet packet decomposition for iris recognition with a Correct Classification Rate (CCR) of 98.375% is proposed. It involves changing the segmentation technique used for this implementation from the integro-differential operator approach (John Daugman's model) to the Hough transform (Wilde's model). This research extensively compared the two segmentation techniques to show which is better in the implementation of the wavelet packet decomposition. Implementation of the integro-differential approach to segmentation showed an accuracy of 91.39% while the Hough Transform approach showed an accuracy of 93.06%. This result indicates that the integration of the Hough Transform into any open source iris recognition module can offer as much as a 1.67% improved accuracy due to improvement in its preprocessing stage. The improved iris segmentation technique using Hough Transform has an overall CCR of 100%.