Finger Knuckle Biometric Authentication using Texture-Based Statistical Approach

Abstract

Reliable authentication techniques play major role in designing security applications. Initially, passwords and tokens are used for identifying the authorized user which are easily forgotten or stolen by others. Therefore, biometric is an evident and secure way of authenticating the user in most of the applications. Various biometric traits such as physical or behavioral characteristics of individual are used for identification. Finger Knuckle Print (FKP) is one of the new emerging modality used for recognition of an individual. Finger Knuckle is rich in texture, a contactless biometric trait with highly unique characteristics. In this work, Gabor filter, texture based statistical approach is used to extract the feature vector from each segmented FKP. In addition, K-nearest neighbor (KNN) algorithm is used to train the system for extracting the feature vector. PolyU database and IIT Delhi databases are used to test the proposed FKP biometric authentication. Proposed FKP biometric authentication technique is implemented and experimental results show that the average efficiency of the algorithm is 86.133% and 99.4% for IIT Delhi and PolyU finger Knuckle databases respectively.