DeepKnuckle: revealing the human identity

Abstract

Identification and authentication are ubiquitous questions which pan across various systems. In certain domains, they are of paramount importance. Like, security forces deploy various human identifications systems to discern potential wrongdoers. They constitute a vital part of various government social welfare schemes. The efficacy of the schemes is greatly impacted by them. Being pervasive and eminent, they demand more dedicated and focused research. Now-a-days, most of the systems incorporate a biometric system to address identification and authentication. The biometric system employs disparate traits like face, signature, iris, fingerprint, palmprint, speech, etc. for identification and authentication. A biometric trait must possess the following fundamental aspects; It should be able to identify an individual uniquely. For an individual, it should be consistent. To acquire it should be easy, cost-effective, time-efficient and automated. On such account, fingerprint trait is of outstanding merit. It has been widely studied and is an integral part of the many present biometric systems. However, fingerprints are subject to occupational hazard. The fingerprint is of abysmal quality for hand labourer, blacksmith, etc. due to the nature of their work. If a fingerprint based biometric system has a large number of such users then its precision is greatly affected. In such scenario, an alternate is to use finger-knuckle-print which possess almost comparable feature as fingerprint while being unaffected by such occupational hazards. In this paper, we propose a novel finger-knuckle-print based biometric system which could be deployed where a large number of user base is rural. Initially, ROI of finger knuckle image has been extracted, enhanced and transformed using the proposed Bubble ordinal pattern (BOP), STAR ordinal pattern (SOP), and Image ray transform (IRT) based locally adapted procedures. A novel DeepMatching technique has been used to perform non-rigid distortion free matching between multiple features of two Finger Knuckle Images (FKI). Finally, the performance of proposed system has been evaluated using score level fusion rule, revealing improvement in the results.