Citation

Ardianto, William (2016) Multi-Instance Finger Vein Recognition System using Local Hybrid Binary Gradient Contour. Masters thesis, Multimedia University.

Full text not available from this repository.

Abstract

Finger vein recognition is a type of biometric technology that uses the vein pattern inside the finger as a personal authenticator. Finger vein images are usually acquired using an infrared imaging device. Finger vein recognition has several advantages over the other hand-based biometrics like fingerprint. For example, finger vein is hard to counterfeit as it reside in the human body, and the vein pattern is not affected by skin condition. Generally, there are three kinds of features that can be extracted from a finger vein, namely line-based, texture-based, and points-based features. Line-based feature uses the blood vascular pattern as a feature, while texture-based feature extracts the texture of the finger vein. On the other hand, points-based feature employs the minutiae of a segmented vein pattern as a feature. Line-based and point-based features can achieve high accuracy in recognition. However the accuracy depends very much on the quality of the image. Low quality images such as optical blurring, irregular shading or noise may lead to false or missing vein pattern, thus affecting the recognition accuracy. To solve this problem, texture-based feature is chosen in this study. This feature representation also has an advantage in terms of processing time because it does not require additional vein pattern extraction process. In this research, Local Hybrid Binary Gradient Contour (LHBGC) and Hierarchical Local Binary Pattern (HLBP) are proposed as the texture descriptor. LHBGC is a modified version of Binary Gradient Contour (BGC). The original BGC only extracts the sign component from the image. However, LHBGC utilises both the sign and magnitude components from the image. Local histogram is used to generate the vector representation from the texture and magnitude components. HLBP is multiscale LBP that fully utilised the non-uniform texture pattern without any training required. HLBP firstly extract the uniform pattern for biggest radius. Then, for those non-uniform patterns, the counterpart LBP of smaller radius is extracted. Among the new LBP, those non-uniform patterns is further processed to extract the uniform pattern in even smaller radius.

Item Type:	Thesis (Masters)
Additional Information:	Call No.: TK7882.B56 A73 2016
Uncontrolled Keywords:	Biometric identification
Subjects:	T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK7800-8360 Electronics
Divisions:	Faculty of Information Science and Technology (FIST)
Depositing User:	Ms Nurul Iqtiani Ahmad
Date Deposited:	07 Sep 2017 10:30
Last Modified:	07 Sep 2017 10:30
URII:	http://shdl.mmu.edu.my/id/eprint/6889

Downloads

Edit (login required)