Personal Verification Based on Finger Knuckle Print Pecognition System Using Raspberry Pi

Abstract

This paper proposes an approach for personal verification based onfinger knuckle recognition (FKR) system using Raspberry Pi 4 andPi camera. We evaluate and compare some advanced methods forbiometric feature extraction and matching in terms of recognitionaccuracy and computational time. From the evaluation and analysis,we found that the Binary Orientation Cooccurrence Vector(BOCV)-which is previously often used for palmprint recognition,is the method of choice for our low-cost and real time FKR system.Along with evaluating our data, we also assess some recent methods on the Hong Kong Polytechnic University Contactless FingerKnuckle Images Database. Experiments show that the BOCVmethod obtains the accuracy over 90% and offers realtime recognition.