Hand vein recognition based on multiple keypoints sets

Vein Biometric Template Security Using Overlapped Shares Accounting for Minimum Storage

“…Coefficients of these curves are the only data that we are storing which is very less in comparison to literature [2,4,8,[11][12][13] where the storage requirement per template is very high. The curves having at least 30 pixels over 100 × 100 windows have been observed to be longest experimentally.…”

“…Raw acquisition of image is succeeded by noise removal using filters like Match, Weiner and Smoothing [2]. Specific noises such as Salt and pepper noise are removed by Gaussian Blurring method.…”

“…Thinning is used for obtaining the skeleton which represents the shape much better and helps in removing all the curved points from the image leaving only the points that relates to the skeleton. Smoothing is done to smooth the contours, fuse narrow breaks and long thin gulfs and remove small holes to obtain the exact structure of the veins [2]. ROI (region of interest) is selected from the acquired images and as it contains lot of information it needs to be preprocessed and features are extracted from it.…”

“…The minutiae points include bifurcation points and ending points [6]. Matching of the templates can be done through phase only correlation [4], Hausdorff distance (HD) is used for matching the two feature sets by computing the distance between every point in both the sets [2], Euclidean distance classifier which is used to check whether it is a genuine vein or imposter vein, Matching using hamming distance where minimum hamming distance from all database templates is find out.…”

“…Fake reference templates may be stored in the data base, Transmission risk is there during enrolment or at the time of acquisition data may be intercepted, etc. [2]. Finger vein system requires small size device so can be used in home, office, car, automobile, etc., for security purpose.…”