INVESTIGATIONS ON DISCRIMINATIVE OPTIMAL FEATURE SELECTION TECHNIQUES FOR HAND BASED BIOMETRIC RECOGNITION

Abstract

Biometric technologies are currently widely used in society, newlinewith applications such as identity and access management, eavesdropping, newlinesecurity systems, social and welfare management, and automatic border newlinecontrol being used directly or indirectly by billions of people. Individuals can newlinebe distinguished using biometrics based on their distinctive physical newlinecharacteristics and behavioural attributes for automated identification newlineverification. newlineThe COVID-19 pandemic, which has been ravaging the planet newlinesince early 2020, is being caused by the novel SARS-Co-V2 coronavirus. newlineThe study focuses on the impact of the COVID-19 pandemic on biometric newlineidentification. Researchers are interested in hand-based contactless newlinebiometrics because they are practical and user-friendly. Consequently, the newlinecost of the device is decreased because both features are generated from a newlinesingle image. There is probably also a decrease in detecting imprecision. This newlinestudy uses contactless finger vein and finger knuckle biometrics to build an newlineeffective biometric recognition system. newlineThe processes that are involved in Finger knuckle Print (FKP) newlineand Finger vein (FV) biometric authentication are feature extraction, fusion, newlinefeature selection, and classification. The proposed two-way multi-algorithm newlinefeature extraction techniques, such as the appearance-based technique and newlinetexture-based algorithm, are used to obtain the feature vectors from the FKP newlineand FV images in order to increase the efficiency of the feature extraction newlinealgorithms. Principal Component Analysis (PCA), Linear Discriminant newlineAnalysis (LDA), and their combinations are proposed for extracting feature newlinevectors for FKP and FV.