Finger Knuckle-Based Multi-Biometric Authentication Systems

Abstract

In the current, digitally revolutionized world, human authentication can be seen as a very important social necessity. All the traditional token or knowledge-based methods alone cannot provide the required level of security. Hence, they are started to be used in conjunction with various biometrics. There are several biometric traits explored till now, such as face, palm, iris, fingerprint, etc., but it has been shown that all of them have several challenges and issues. Recently, hand-based biometric traits such as palm, fingerprint, knuckle, and vein patterns have started to get huge amount of attention due to their easy and inexpensive acquisition and better performance.

In this chapter, several state-of-the-art finger knuckle print-based authentication systems have been discussed. Many recently proposed algorithms, for extracting region of interest (ROI) with the help of curvature Gabor filters or convex coding, have been discussed. Since image quality plays a significant role, several finger knuckle trait-based image quality parameters have been described. Various finger knuckle ROI enhancement procedures have been highlighted along with multiple feature extraction and matching algorithms. All systems have been tested on few publicly available finger knuckle print/image databases such as PolyU and IITD datasets and are compared using the standard performance parameters such as equal error rate (EER) and correct recognition rate (CRR).