The Finger-Knuckle-Print recognition using the kernel principal components analysis and the support vector machines

Abstract

In the computer networks explosion's time, the need to identify individuals increasingly becomes necessary to perform various operations, such as access control and secure payments. So far, inputting alphanumeric code remains the most used solution. This solution, in spite of having the merit to be very simple, has the disadvantage to certify only the individual who enters the correct code. Another possibility that is open to us is to use biometric identification, by identifying directly the physical traits of the user. Biometric identification is defined as a science allowing the identification of people using their behavioral or physiologic characteristics. It seems like an evident solution to the problem explained previously: the identity of a person is then related to who he/she is and not to what he/she possesses or knows. In this work, we propose a biometric system based on a very recent biometric trait, which consists in the finger-Knuckle-Prints. This recognition is based on a mathematical model.