BIOMETRIC RECOGNITION

Abstract

Biometric Recognition is the essential process of authenticating an individual and has a very wide application area starting from one’s essential need like phone or laptop access to high end applications like in, Airport, Border control, surveillance, forensic applications etc. It is implemented almost in every system that require some sort of authentication for establishing the identity of a person. Biometrics recognition is based on the Biometric traits, which are the physical or behavioral characteristics of a person that are physically or behaviorally linked to the person. Since Biometric traits are physically linked to the user therefore very difficult to steal or forge and person do not require to remember the login or password for accessing any system or premises. There are various biometric traits like Face, Finger Print, Finger veins, Iris, Scalera, etc which are extensively utilized in several recognition applications. Some recognition applications use number of traits of a person to have high recognition accuracy. Some even take physical as well as behavioral along with soft biometric traits like hand shape etc. to have robust recognition system that works in unconstrained environment with higher accuracy. Purpose of this research work is to improve the accuracy of the recognition system by taking number of traits of a person together called Multi-Modal Biometric (MMB) Recognition and by focusing on the finger vein trait, which is one of the current research areas as it can be captured only of a live person. A new method is proposed, the MMB recognition system centered on the Features Level and Scores Level (FLSL) fusion method and Modified Deep Learning Neural Network (MDLNN) classifier in order to enhance the performance. The face, ear, retina, fingerprint, and front hand image traits are considered by the proposed method.

iii The unique patterns of finger veins (FV) are utilized by Finger vein recognition (FVR) for detecting individuals at a high-level accuracy. However, on account of the existence of artifacts, irregular shading, distortions, etc, precise FV detection is a difficult task. A framework for identifying FV is created by the work to offer a precise biometric authorization utilizing Enhanced Sigmoid Reweighted based Convolutional Neural Network (ES-RwCNN).