IKP based human authentication using multi-feature fusion

To develop a biometric system using finger knuckle patterns has recently grown interest of researchers worldwide. Compared with its hand counterparts, skin patterns lie nearby the joints of finger surface has several unique merits. Knuckle print based biometric system is highly acceptable because these patterns do not disturb with injury and wound on the finger surface. However, geometrical and textural features have been used widely in most of the existing knuckle print recognition methods where an individual is verified by extraction of lines, creases, and shape features. This paper presents a score level fusion scheme for personal recognition using novel inner knuckle print based features i.e, knuckle distance based geometrical features, and local direction pattern based texture features. The proposed method is tested on our in-house Hand database (720 images) collected over a span of 3 months. Finally, the test and training images are compared in terms of using a K-NN classifier. Experimental results reveal the performance of the proposed FKP system in terms of CRR (96%), EER (3.45%), and computation time.