Multimodal approach for hand biometrics

Abstract

Biometrics is the technique to verify identify a person using his physiological and or behavioural trait Biometric systems use trait which is universal unique collectable and user friendly Unimodal biometric systems use single trait of a person Unimodal system faces problems due to illness scar in finger throat infection non universality of selected trait spoof attack and if intraclass variations are more and interclass variance is less To avoid these problems multimodal biometric algorithms are developed by many researchers and are adopted in many organizations It combines two or more traits to get better results The fusion of traits can be done at representation feature level matching score level or decision level newlineThis work is based on designing a hand based biometric system It has been chosen because of cost effectiveness simplicity user friendliness and less storage space requirement It can be used for verification purpose or identification purpose In verification mode system has to verify the claimed user which is done by one to one matching and in identification mode system has to identify the user from the complete database which is done by one to many matching newlineA hand based biometric system consists of basically five modules namely 1 Image acquisition module 2 Image Preprocessing module 3 Feature Extraction Module 4 Matching Module and 5 Decision module newlineIn literature hand based biometric systems are largely classified on the basis of acquisition and the set of features used for generating template On the basis of image acquisition hand biometric system can be classified as peg based peg free contact based and contact free system Pegs are the guiding pins present on the platform to help the user in properly placing the hand on platform In peg free systems user is free to place his hand on platform In contactless environment there is no need for placing hand User has to keep his hand in air and photograph of hand is taken from a particular distance using camera Peg based systems are costly