TECHNOLOGY Human Identification Based On Scleral Vasculature
Rujuta Makarand Vaidya, Sarika Patil · 2014
The blood vessel patterns of sclera show rich and u nique details. To date, sclera is relatively less s tudied and little is known regarding its usefulness and pe rformance. Here, sclera recognition for human ident ification is proposed. It poses a challenge because sclera vesse l patterns are often dense and/or saturated and, mo st importantly, the vessel structure in the sclera is complex and h as deformations. A new method for sclera segmentation which uses Daugman’s integro-differential operator is proposed . It is followed by Otsu’s thresholding method and erosion. A comparative analysis is presented between Gabor fil ter and Canny edge detection methods. Since, Canny filter has higher noise immunity as compared to Gabor filter i t is used to enhance and thin the conjunctival vasc ulature patterns. A sclera template generation method is pr oposed that uses Hough transform. The proposed scheme is tested on the images from UBIRIS 3 database which is publicly available. Thus, the use of sclera vasculature pattern as potential biometric is investigated.