A New Approach to Vein Pattern Authentication

Abstract

This paper presents a new approach to vein pattern authentication system. The various stages involved are binarisation of the ROI, thinning, feature extraction  and matching. Our main focus is on Dynamic ROI Extraction and thinning. We got a database of hand images which are far infrared images of hand. For that images, we have done dynamic ROI extraction. The advantage of dynamic ROI extraction is that ROI extracted for different hand images varies in size as the size of the hand varies. As the size of ROI varies, we can extract more features from larger hand. We’ve proposed a new thinning algorithm which involves thinning the ROI to one pixel thick medial axis vein network and also removal of thinning artifacts. We used a sequential thinning algorithm for thinning. We used a flag map in addition to image to decide which pixel to delete during thinning process. An artifact removal algorithm is used to remove any extra artifacts which ends the thinning process. After that key features are extracted which are bifurcations and end points which are stored separately as templates. Finally a matching algorithm is applied for identification. The matching algorithm is a coordinate matching where in we match the coordinate positions of key features of the live image which are stored

in live template with the stored templates with certain fixed threshold. Our system is efficient as it got the lowest FAR which is necessary for security.