¹Department of Computer Science and Engineering, India.
²Annamalai Univeristy, Annamalainagar, Tamil Nadu, India.

---

Abstract—Nowadays a biometric recognition system placed an important role in individual identification and authentication process. The user identities are authenticated by utilizing different biometric such as ear, face, eye, iris, palm, signature, finger knuckle and so on. From the various biometric features, finger-knuckle, having many advantages like the user acceptance of the outer-palm surface is very high, rich in texture feature, easily accessible, stable features because the features are never changed during the person emotionally and behavioral aspects. So, the proposed system uses the finger knuckle based biometric system for authenticating the user identities. Initially the captured biometric images are converted to the grayscale format and the noise has been removed by applying the Non-Local Median filter which improves the quality of the captured image. From the preprocessed image the contour has been extracted by using the Graph based Geometric Approach and the Principal Curvature based finger-knuckle-print Region has been located. The located key point is trained by Compositional Neural Networks and the matching is done with the help of the Levenshtein distance measure which determines whether an individual is authorized or not. Then the proposed system is implemented using PollyU finger knuckle database and the efficiency is analyzed in terms of false acceptance rate, false rejection rate and equal error rate.

Index Terms—Biometric recognition, grayscale image, non-local median filter, graph based geometric approach, principal curvature based region, levenshtein distance measure.

[PDF]

Cite: S. Suganthi Devi, A. Suhasini, "Graph Geometric Approach and Bow Region Based Finger Knuckle Biometric Identification System," Journal of Software vol. 12, no. 9, pp. 682-694, 2017.