FMK (Finger Major Knuckle)

The recognition of persons is based on what he knows (PIN, password), what he has (cards) and what he is (biometric characteristics). However, biometrics such as fingerprint and face are more secure than traditional methods. In fact, people can be recognized by a lot of unique biometric features either physiological or behavioral. Therefore, a biometric security method for a smartphone is proposed in this paper for easier use and to improve recognition results by using a new knuckle modality: the major knuckle of the thumb finger modality instead of the major knuckle of the middle finger modality which is the most used in related works. In finger knuckle recognition domains, different CNN (Convolutional Neural Network) architectures have been applied, requiring a huge database. To deal with this problem, this paper presents a Siamese network for identity verification using contactless major knuckle of thumb fingerprint modality. Actually, the Siamese network has been well used for one-shot learning that aims at learning and recognizing from little data. The FMK (Finger Major Knuckle) dataset was proposed and constructed for testing and evaluation. In fact, this study achieved good results with accuracy of 81.37% for the major knuckle of the left thumb finger and 79.62% for the major knuckle of the left middle finger.