Back to results list
Show full item record
Please use this identifier to cite or link to this item:
|Title:||Personal authentication using Finger-Knuckle-Print||Authors:||Zhang, Lin||Degree:||Ph.D.||Issue Date:||2011||Abstract:||Biometrics, which is the discipline of recognizing a person's identity based on his/her physical or behavioural characteristics, has attracted much attention in the recent decade due to its numerous applications. Each traditional biometric identifier has its own advantages and disadvantages. Thus, researchers have never stopped searching for new kinds of biometric identifiers. By observing that the texture pattern produced by bending the finger knuckle is highly distinctive, in this work we present a new biometric authentication system using finger-knuckle-print (FKP) imaging. Various aspects of this system are thoroughly investigated and discussed in this thesis. Finger-knuckle-print, which refers to the inherent skin patterns of the outer surface around the phalangeal joint of one's finger, is a new member of the biometrics family. It has high capability to discriminate different individuals. As a kind of hand-based biometrics, FKP has the merit of high user friendliness. Compared with the other traditional hand-based biometric identifiers, for example the fingerprint, FKP may have some special advantages. At first, it is not easy to be abraded and forged since people usually hold stuffs with the inner side of the hand. Moreover, unlike the fingerprint, there is no stigma of criminal investigation associated with the FKP, so it can have a high user acceptance. These characteristics make the FKP have a great potential to be a widely accepted promising biometric identifier. Our FKP recognition system comprises four major components: FKP image acquisition, ROI (region of interest) extraction, feature extraction, and feature matching. At first, a specially designed FKP image acquisition device is established. It is composed of a finger bracket, a ring LED light source, a lens, and a CCD camera. Such a device can be conveniently used and can capture high quality FKP images. With the developed acquisition device, a large FKP database containing 7920 samples collected from 660 fingers is established and now it is publicly available.
After an FKP image is captured, a region of interest (ROI) needs to be cropped from the original image for the further feature extraction and matching. Such a pre-processing step can reduce the data amount in the feature extraction and matching stage and can abate the influence of variations of the FKPs. To this end, an efficient FKP ROI extraction algorithm is proposed based on the intrinsic characteristics of FKP images. As in any pattern classification task, the feature extraction and matching plays a key role in our FKP-based personal authentication system. To this end, we have developed and examined a couple of different methods. At first, the performances of several state-of-the-art coding based methods are evaluated, including CompCode, RLOC, and OrdinalCode. After that, a novel coding-based feature extraction and matching method, namely ImCompCode&MagCode is proposed, which is an extended version of CompCode. Moreover, we propose two more efficient and compact coding based feature extraction and matching approaches, RCode1 and RCode2, using Riesz transforms. In fact, coding based feature extraction and matching methods make use of local information of images. Actually, global information hidden in images can also be exploited for recognition. Based on this belief, we have developed another FKP matching method by matching the Fourier transform coefficients of two FKP images using the phase-only correlation technique. Furthermore, based on the results of psychophysics and neurophysiology studies that both local and global information is crucial for the image perception, we present an effective FKP recognition scheme by extracting and assembling local and global features of FKP images. Specifically, we use the local orientation, the local phase, and the local phase congruency as the local features and they can be extracted by using Gabor filters. By increasing the scale of Gabor filters to infinity, actually we get the Fourier transform of the image, and hence the Fourier transform coefficients of the image can be taken as the global features. Such kinds of local and global features are naturally linked via the framework of the time-frequency analysis. All the developed recognition methods are thoroughly investigated on our established benchmark FKP database. And it needs to be noted that the feature extraction and matching methods developed in this thesis can also be applied to some other biometrics systems, e.g., the palmprint recognition system. Another contribution of this thesis is that the developed technologies have been implemented in a standalone embedded FKP recognition system, which can be readily used in practice. Such a system is the first of its kind.
Pattern recognition systems.
Hong Kong Polytechnic University -- Dissertations
|Pages:||xiv, 112 p. : ill. ; 30 cm.|
|Appears in Collections:||Thesis|
View full-text via https://theses.lib.polyu.edu.hk/handle/200/6208
Citations as of May 22, 2022
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.