Finger-Knuckle-Print ROI Extraction using Curvature Gabor Filter for

Human Authentication

Aditya Nigam

and Phalguni Gupta

School of Computer Science and Electrical Engineering, Indian Institute of Technology Mandi (IIT Mandi), Mandi, India

National Institute of Technical Teacher’s & Research (NITTTR), Salt Lake, Kolkata, India

Keywords:

Finger-Knuckle-Print, FKP, Curvature Gabor Filter, Segmentation, Authentication.

Abstract:

Biometric based human recognition is a most obvious method for automatically resolving personal identity

with high reliability. In this paper we present a novel ﬁnger-knuckle-print ROI extraction algorithm. The

basic Gabor ﬁlter is modiﬁed to Curvature Gabor Filter (CGF) to obtain central knuckle line and central

knuckle point which are further used to extract FKP ROI image. Largest public FKP database is used for

testing which consists of 7, 920 images collected from 660 different ﬁngers. The results has been compared

with the only other existing Convex Direction Coding (CDC) ROI extraction algorithm. It has been observed

that the proposed algorithm achieves better performance with EER drop percentage more than 20% in all

experiments. This suggests that the proposed CGF algorithm has been extracting ROI more consistently then

CDC and hence can facilitates any ﬁnger-knuckle-print based biometric systems.

1 INTRODUCTION

Biometric based authentication system has been used

widely in commercial and law enforcement applica-

tions. The use of various biometric traits such as

ﬁngerprint, face, iris, ear, palmprint, hand geometry

and voice has been well studied (Jain et al., 2007).

Recently some multimodal systems has also been

reported in (Nigam and Gupta, 2015a; Nigam and

Gupta, 2014a; Nigam and Gupta, 2013a) fusing vari-

ous combinations of palmprint, knuckleprint and iris

images in pursuit of superior performance. The qual-

ity estimation of any biometric trait is also a very im-

portant and difﬁcult task because quality is directly

proportional to system performance. Good amount of

work has been done to estimate the quality of face

and ﬁngerprint images due to the presence of some

very speciﬁc texture. But limited work is done so far

for iris (Nigam et al., 2013), knuckleprint (Nigam and

Gupta, 2013b) and palmprint quality estimation as its

lacks any such speciﬁc texture and structure. Some

more work on knuckleprint and palmprint recogni-

tion is reported in (Badrinath et al., 2011; Nigam

and Gupta, 2011; Nigam and Gupta, 2014b) using

SIFT and SURF fusion and LK−tracking of corner

features.

It is reported that the skin pattern on the ﬁnger-

knuckle is highly rich in texture due to skin folds and

creases, and hence, can be considered as a biometric

identiﬁer (Woodard and Flynn, 2005).

Figure 1: Finger-Knuckle-Print Anatomy.

1.1 Motivation

The line like (i.e. knuckle lines) rich pattern struc-

tures in vertical as well as horizontal directions exist

over ﬁnger-knuckle-print, as shown in Fig. 1. These

horizontal and vertical pattern formations are believed

to be very discriminative (Zhang et al., 2011a). The

ﬁnger-knuckle-print texture is developed very early

and last very long because it occurs on the outer side

of the hand and no one can use them for almost any

work except boxers. Negligible wear and tear as well

as print quality degradation with time and age are ob-

served. Its failure to enrollment rate is also expected

to be very low as compared to the ﬁngerprint and it

does not require much user cooperation. Further, ad-

366

Nigam, A. and Gupta, P.

Finger-Knuckle-Print ROI Extraction using Curvature Gabor Filter for Human Authentication.

DOI: 10.5220/0005724103640371

In Proceedings of the 11th Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2016) - Volume 3: VISAPP, pages 366-373

ISBN: 978-989-758-175-5

Figure 2: Fingerprint Vs FKP (Row 1 shows ﬁngerprints while second row shows the corresponding knuckleprints).

vantages of using FKP include rich in texture features

(ChorasÂt’ and and Kozik, 2010), easily accessible,

contact-less image acquisition, invariant to emotions

and other behavioral aspects such as tiredness, stable

features (Zhang et al., 2011b) and acceptability in the

society (Kumar and Zhou, 2009).

1.2 FKP Vs Fingerprint

In this work we have considered knuckleprint over

ﬁngerprint mainly because it is observed that in ru-

ral areas the ﬁngerprint quality is very poor. The

cultivators and hard workers use their hands very

roughly, causing sever damage to their ﬁngerprints

permanently. But still there knuckleprint quality is

very good because no one can use them for any work.

Hence it can be inferred that rural knuckleprint qual-

ity is better than rural ﬁngerprint as one can also ob-

serve from Fig. 2. In such a scenario it has low failure

to enroll rate FT E as compared to ﬁngerprint and can

be easily acquired using an inexpensive setup with

lesser user cooperation. Also user acceptance favors

knuckleprint as unlike ﬁngerprint it has never being

associated to any criminal investigations in the past.

1.3 Literature Review

Despite of these characteristics and advantages of us-

ing FKP as biometric identiﬁer, limited work has been

reported in the literature (Jungbluth, 1989) as it is a

relatively new biometric trait. Systems reported in lit-

erature have used global features, local features and

their combinations (Zhang et al., 2011b) to repre-

sent FKP images. In (Kumar and Ravikanth, 2009),

FKP features are extracted using principle compo-

nent analysis (PCA), independent component anal-

ysis (ICA) and linear discriminant analysis (LDA).

In (Zhang et al., 2009b), FKP is transformed using

Fourier transform and band-limited phase only corre-

lation (BLPOC) is employed to match the FKP im-

Figure 3: Overall architecture of FKP recognition system.

ages. In (Zhang et al., 2009a), Zhang et.al extracted

ROI using convex direction coding and used BLPOC

for matching. In (Morales et al., 2011a), knuck-

leprints are enhanced using CLAHE and SIFT key-

points are used for matching. In (Xiong et al., 2011),

features are extracted using local gabor binary pat-

terns and a discriminating local pattern is extracted

to represent that pixel. In (Zhang et al., 2011a), lo-

cal gabor and global BLPOC features are fused to

achieve better performance. Both local and global

scores are fused to get the better result. In (Zhang

et al., 2010b), a bank of six gabor ﬁlters at an angle

is applied to extract features for those pixels that

are having varying gabor responses. In (Zhang et al.,

2012), three local features viz. phase congruency, lo-

cal orientation and local phase are fused at score level.

Global feature can extract the general appearance

(holistic characteristics) of FKP which is suitable for

coarse level representation, while local feature pro-

vides more detailed information for any speciﬁc lo-

cal region which is appropriate for ﬁner representa-

Finger-Knuckle-Print ROI Extraction using Curvature Gabor Filter for Human Authentication

367

(a) Raw ﬁnger-knuckle-print (b) Annotated FKP (c) FKP ROI

Figure 4: Finger-knuckle-print ROI Annotation.

tion (Zhang et al., 2011b). There exist systems where

local features of FKP are extracted using the Ga-

bor ﬁlter based competitive code (CompCode) (Zhang

and Zhang, 2009) and magnitude information (Im-

CompCode&MagCode) (Zhang et al., 2010c). Fur-

ther, in (Kumar and Zhou, 2009), orientation of ran-

dom knuckle lines and crease points (KnuckleCodes)

of FKP which are determined using radon transform

are used as features. In (Woodard and Flynn, 2005),

FKP is represented by curvature based shape index.

Morales et. al., (Morales et al., 2011b) have proposed

an FKP based authentication system (OE-SIFT) using

scale invariant feature transform (SIFT) from orien-

tation enhanced FKP. In (ChorasÂt’ and and Kozik,

2010), an hierarchical veriﬁcation system using prob-

abilistic hough transform (PHT) for coarse level clas-

siﬁcation and the speeded up robust features (SURF)

for ﬁner classiﬁcation has been proposed. SIFT and

SURF features of FKP are matched using similarity

threshold (Morales et al., 2011b). In (Zhang et al.,

2010a), features are extracted using Hilbert transform

(MonogenicCode). Further, Zhang et.al (Zhang et al.,

2011b) have proposed a veriﬁcation system which is

by BLPOC (Zhang et al., 2009b) and the local in-

formation obtained by Compcode (Zhang and Zhang,

2009). However, there does not exist any system

which is robust to scale and rotation.

The ROI extraction is an initial phase and is very

crucial because all other modules has been using its

output. Incorrect ROI extraction render all other steps

meaningless. This paper deals with the problem of

designing an efﬁcient ﬁnger-knuckle-print ROI ex-

traction system. It is compared with the only other

available algorithm Convex Direction Coding (CDC)

(Zhang et al., 2011a; Zhang et al., 2010c). The ﬁnger-

knuckle-print ROI is extracted by applying a mod-

iﬁed version of gabor ﬁlter to estimate the central

knuckle line and point as shown in Fig. 4(b). The

central knuckle point is used to extract consistently

the ﬁnger-knuckle-print ROI from any image.

Any FKP system consists of ﬁve major tasks,

viz. ROI extraction, quality estimation, ROI pre-

processing, feature extraction and matching. The

overall architecture of any ﬁnger-knuckle-print based

recognition system is shown in Fig. 3. The pub-

licly available ﬁnger-knuckle-print PolyU database

(PolyU, 2010) is used to test the proposed system that

contains both the raw sensor images (that are used

to segment) as well as extracted ROI (that are used

to compared performance). The PolyU database con-

tains images that are acquired using normal web-cam

of resolution 384× 288 using their indigenous captur-

ing device. The device allows the user to place only

one ﬁnger at a time and acquires images that are hori-

zontally aligned. Therefore, at the time of ROI extrac-

tion, raw ﬁnger-knuckle-print images are assumed to

be horizontal and contain single ﬁnger knuckle.

2 PROPOSED FKP ROI

EXTRACTION

This section proposes an efﬁcient ﬁnger-knuckle-

print ROI extraction technique. The prime objective

of any ROI extraction technique is to segment same

region of interest consistently from all images. The

central knuckle point as shown in Figure 4(b) can be

used to segment any ﬁnger-knuckle-print consistently.

Since ﬁnger-knuckle-print is aligned horizontally, one

can now easily extract the central region of interest

from any ﬁnger-knuckle-print that contains rich and

discriminative texture using this point. The proposed

ROI extraction algorithm performs in three steps; de-

tection of knuckle area, central knuckle-line and cen-

tral knuckle-point deﬁned as follows.

2.1 Knuckle Area Detection

In this step, the whole knuckle area is segmented from

the background in order to discard background region.

The acquired ﬁnger-knuckle-print may be of poor

quality. Hence, each ﬁnger-knuckle-print is enhanced

using contrast limited adaptive histogram equaliza-

tion (Pizer et al., 1987) (CLAHE) to obtain better

VISAPP 2016 - International Conference on Computer Vision Theory and Applications

368

(a) Raw (b) Binary (c) Canny (d) Boundary (e) FKP Area

Figure 5: Steps involved in Knuckle Area Detection.

edge representation which helps to detect the knuckle

area more robustly. CLAHE divides the whole image

into blocks of size 8× 8 and applies histogram equal-

ization over each block. The enhanced image is bina-

rized using Otsu thresholding that segments the image

into two clusters (knuckle region and background re-

gion) based on their gray values. Such a binary image

is shown in Fig. 5(b). It can be observed that the

knuckle region may not be accurate because of sensor

noise and background clutter. This can be obtained

by using canny edge detection. A resultant image is

shown in Fig. 5(c) and the largest connected compo-

nent is considered as the required knuckle boundary.

The detected boundary is eroded to smooth it and re-

move any discontinuity as shown in Fig. 5(d). Fi-

nally all pixels within the convex hull of the knuckle

boundary are considered as the knuckle area. Figure

5(e) shows a segmented knuckle area. Some top and

bottom rows are assumed to be background and are

discarded from the raw image.

2.2 Central Knuckle-Line Detection

The central knuckle line is deﬁned as that column

of the image with respect to which the knuckle can

be considered as symmetric as observed from Figure

4(b). This line is used to extract the ﬁnger-knuckle-

print ROI. A very speciﬁc and symmetric texture is

observed around the central knuckle line which is

used for its detection. To perceive such a speciﬁc

texture, a knuckle ﬁlter is created by modifying the

conventional gabor ﬁlter which is deﬁned as follows.

[A] Knuckle Filter: The conventional gabor ﬁl-

ter is created when a complex sinusoid is multiplied

with a Gaussian envelope as deﬁned in Eq .(1) and is

shown in Figure 6(f).

G(x, y;γ, θ, ψ, λ, σ) = e

−(

·γ

2·σ

)

| {z }

Gaussian Envelope

× e

2πX

+ψ)

| {z }

Complex Sinusoid

(1)

where x and y are the spatial co-ordinates of the ﬁlter

and X , Y are obtained by rotating x, y by an angle θ

using the following equations:

X = x ∗ cos(θ) + y ∗ sin(θ) (2)

Y = −x ∗ sin(θ) + y ∗ cos(θ) (3)

In order to model the curved convex knuckle lines,

a knuckle ﬁlter is obtained by introducing curvature

parameter in the conventional gabor ﬁlter. The basic

gabor ﬁlter equation remains to be the same (as in Eq.

(1)). Only X and Y co-ordinates are modiﬁed as fol-

lows:

X = x ∗ cos(θ) + y ∗ sin(θ) + c ∗ (−x ∗ sin(θ) + y ∗ cos(θ))

(4)

Y = −x∗ sin(θ)+y∗ cos(θ) (5)

The curvature of the gabor ﬁlter can be modulated

by the curvature parameter. This curved gabor ﬁlter

with parameters (γ = 1, θ = π, ψ = 1, λ = 20, σ = 20)

can be used for knuckle ﬁlter creation. The value of

curvature parameter is varied as shown in Fig. 8 and

its optimal value for our database is selected heuristi-

cally. The proposed knuckle ﬁlter is obtained by con-

catenating two such curved gabor ﬁlters ( f

, f

) en-

suring the distance between them as d. The ﬁrst ﬁlter

( f

) is obtained using the above mentioned parame-

ters while the second ﬁlter (f

= f

f lip

) is just the ver-

tically ﬂipped version of the ﬁrst ﬁlter because ﬁnger-

knuckle-prints are vertically symmetric. In Figure 8,

several knuckle ﬁlters are shown with varying cur-

vature and distance parameters. One can observe

that increasing curvature parameter c introduces more

and more curvature in the ﬁlter. Finally c = 0.01

and d = 30 are considered for selected knuckle ﬁlter

0.01,30

[B] Knuckle Line Extraction: All pixels

belonging to knuckle area are convolved with the

knuckle ﬁlter F

0.01,30

. Pixels over the central knuckle

line must be having higher response as compared to

others because of ﬁlter’s shape and construction. The

ﬁlter response for each pixel is binarized using thresh-

old as f ∗ max where max is the maximum knuckle ﬁl-

ter response and f ∈ 0 to 1 is a fractional value. The

Finger-Knuckle-Print ROI Extraction using Curvature Gabor Filter for Human Authentication

369

(a) 1D Gaussian (b) 1D Sinusoid (c) 1D Gabor

(d) 2D Gaussian (e) 2D Sinusoid (f) 2D Gabor (g) 2D Gabor Filter

Figure 6: Conventional 1 and 2 Dimensional Gabor Filter (Prasad and Domke, 2007).

binarized ﬁlter response is shown in Fig. 9(a) where

it is super imposed over knuckle area with blue color.

The column-wise sum of the ﬁlter response for each

column is computed. The central knuckle line is con-

sidered as, that column which is having the maximum

knuckle ﬁlter response as shown in Fig. 9(b).

(a) Curved Knuckle Lines

(b) Curvature Knuckle Filter

Figure 7: Curvature Knuckle Filter.

2.3 Central Knuckle-Point Detection

The central knuckle point is required to crop the

knuckle-print ROI that must lie over central knuckle

line. Hence the top and the bottom point over central

knuckle line, belonging to the knuckle area, are com-

puted and their mid-point is considered as the central

knuckle point. It is shown in Figure 9(b). The re-

quired ﬁnger-knuckle-print ROI is extracted as a re-

gion of size (2∗ w + 1) × (2 ∗ h + 1) considering cen-

(a) c=0.00,d=0 (b) c=0.00,d=30

(e) c=0.04,d=0 (f) c=0.04,d=30

Figure 8: Various Knuckle Filter Arrangements.

tral knuckle point as the center. It is shown in Fig.

9(c).

Given a raw ﬁnger-knuckle-print image I, Algo-

rithm 1 can be used to extract the ﬁnger-knuckle-print

ROI (FKP

ROI

) from it.

3 EXPERIMENTAL ANALYSIS

The proposed ROI extraction algorithm is tested

over largest publicly available ﬁnger-knuckle-print

database (PolyU, 2010). Some of the segmented

PolyU ﬁnger-knuckle-print database images are

shown in Fig. 10. One can observe that the proposed

algorithm can extract consistently the ROI of images

VISAPP 2016 - International Conference on Computer Vision Theory and Applications

370

(a) Knuckle Filter Response (b) Annotated (c) FKP ROI

Figure 9: Finger-knuckle-print ROI detection. (a) Knuckle ﬁlter response is super impose over the knuckle area with blue

color, (b) Full Annotated and (c) FKP (FKP

ROI

in the database. The correct segmentation accuracy of

the proposed algorithm is observed as 95.15% =

7536

7920

over the PolyU ﬁnger-knuckle-print (PolyU, 2010)

database that contains images acquired from 660 sub-

jects. Some images for which the proposed algorithm

failed to segment are shown in Fig. 11. The algorithm

fails mainly due to poor image quality. Finally such

subjects are segmented manually.

Algorithm 1: Finger-knuckle-print ROI Detection.

Require:

Raw ﬁnger-knuckle-print image I of size m × n.

Ensure:

The ﬁnger-knuckle-print ROI FKP

ROI

, of size

(2 ∗ w + 1) × (2 ∗ h + 1).

1: Enhance the FKP image I to I

using CLAHE;

2: Binarize I

to I

using Otsu thresholding;

3: Apply Canny edge detection over I

to get I

cedges

;

4: Extract the largest connected component in I

cedges

as FKP raw boundary, (FKP

raw

Bound

);

5: Erode the detected boundary FKP

raw

Bound

to ob-

tain continuous and smooth FKP boundary,

FKP

smooth

Bound

;

6: Extract the knuckle area K

= All pixels in image

I ∈ the ConvexHull(FKP

smooth

Bound

);

7: Apply the knuckle ﬁlter F

0.01,30

over all pixels

∈ K

;

8: Binarize the ﬁlter response using f ∗ max as the

threshold;

9: The central knuckle line (c

), is assigned as that

column which is having the maximum knuckle

ﬁlter response;

10: The mid-point of top and bottom boundary points

over c

∈ K

, is deﬁned as the central knuckle

point (c

11: The knuckle ROI (FKP

ROI

) is extracted as the re-

gion of size (2 ∗ w + 1) × (2 ∗ h + 1) from raw

ﬁnger-knuckle-print image I, considering c

its center point.

The only other available ROI extraction algorithm

is based on Convex Direction Coding (CDC) (Zhang

(a) Subject A

(b) Subject B

Figure 10: Some Correctly Segmented ﬁnger-knuckle-

prints.

Figure 11: Failed Knuckle ROI detection.

et al., 2011a). The proposed Curvature Gabor Fil-

ter (CGF) based algorithm is compared with CDC as

shown in Fig 12 and Table 1. The images are cropped

using both methods and are matched using algorithm

as proposed in (Nigam and Gupta, 2015b). Exactly

same testing protocol is used in order to make a fair

comparison. One can clearly observe that all perfor-

mance parameters are improved for all ﬁnger cate-

gories as well as over full FKP database. In all cases

EER drop

percentage was more than 20% suggest-

ing that the proposed CGF algorithm has been ex-

tracting ROI more consistently.

4 CONCLUSION

The ROI extraction can be considered as the most

crucial stage in any recognition system. In this pa-

per a novel ﬁnger-knuckle-print ROI extraction algo-

rithm is presented and compared with the only other

available algorithm Convex Direction Coding (CDC)

(Zhang et al., 2011a; Zhang et al., 2010c) over PolyU

FKP database. The ﬁnger-knuckle-print ROI is ex-

tracted by applying a modiﬁed version of gabor ﬁlter

% drop is calculated with respect to CDC

Finger-Knuckle-Print ROI Extraction using Curvature Gabor Filter for Human Authentication

371

Figure 12: Comparative Analysis using same matching algorithm based on CIOF (Nigam and Gupta, 2015b), while ROI

segmentation is done using CDC and CGF.

Table 1: Comparative Analysis : Curvature Gabor Filter (CGF) Vs Convex Direction Coding (CDC) (Zhang et al., 2011a).

Db DI EER Accuracy EUC CRR

CGF CDC CGF CDC Drop(%) CGF CDC CGF CDC CGF CDC

FKP 1.65 1.63 0.71 0.89 20.22 99.38 99.28 0.165 0.201 99.97 99.94

LI 1.63 1.63 0.858 1.077 20.33 99.25 99.15 0.228 0.302 100 99.79

LM 1.68 1.68 0.791 1.009 21.6 99.32 99.19 0.298 0.356 100 100

RI 1.61 1.61 0.58 0.740 21.6 99.46 99.37 0.068 0.086 100 100

RM 1.67 1.66 0.875 1.094 20.01 99.25 99.2 0.213 0.265 100 100

to estimate the central knuckle line and point. It is

observed that in all experiments EER drop percent-

age was more than 20% suggesting that the proposed

CGF algorithm has been extracting ROI more consis-

tently then CDC.

REFERENCES

Badrinath, G., Nigam, A., and Gupta, P. (2011). An efﬁcient

ﬁnger-knuckle-print based recognition system fusing

sift and surf matching scores. In Qing, S., Susilo, W.,

Wang, G., and Liu, D., editors, Information and Com-

munications Security, volume 7043 of Lecture Notes

in Computer Science, pages 374–387. Springer Berlin

Heidelberg.

ChorasÂt’ and, M. and Kozik, R. (2010). Knuckle bio-

metrics based on texture features. In International

Workshop on Emerging Techniques and Challenges

for Hand-Based Biometrics, pages 1–5.

Jain, A. K., Flynn, P., and Ross, A. A. (2007). Handbook of

Biometrics. Springer-Verlag, USA.

Jungbluth, W. O. (1989). Knuckle print identiﬁcation. Jour-

nal of Forensic Identiﬁcation, 39:375–380.

Kumar, A. and Ravikanth, C. (2009). Personal authentica-

tion using ﬁnger knuckle surface. IEEE Transactions

on Information Forensics and Security, 4(1):98 –110.

Kumar, A. and Zhou, Y. (2009). Personal identiﬁcation us-

ing ﬁnger knuckle orientation features. Electronics

Letters, 45(20):1023 –1025.

Morales, A., Travieso, C., Ferrer, M., and Alonso, J.

(2011a). Improved ﬁnger-knuckle-print authentica-

tion based on orientation enhancement. Electronics

Letters, 47(6):380–381.

Morales, A., Travieso, C., Ferrer, M., and Alonso, J.

(2011b). Improved ﬁnger-knuckle-print authentica-

tion based on orientation enhancement. Electronics

Letters, 47(6):380 –381.

Nigam, A. and Gupta, P. (2011). Finger knuckleprint based

recognition system using feature tracking. In Sun,

Z., Lai, J., Chen, X., and Tan, T., editors, Biometric

VISAPP 2016 - International Conference on Computer Vision Theory and Applications

372

Recognition, volume 7098 of Lecture Notes in Com-

puter Science, pages 125–132. Springer Berlin Hei-

delberg.

Nigam, A. and Gupta, P. (2013a). Multimodal personal

authentication system fusing palmprint and knuck-

leprint. In Huang, D.-S., Gupta, P., Wang, L., and

Gromiha, M., editors, Emerging Intelligent Com-

puting Technology and Applications, volume 375 of

Communications in Computer and Information Sci-

ence, pages 188–193. Springer Berlin Heidelberg.

Nigam, A. and Gupta, P. (2013b). Quality assessment of

knuckleprint biometric images. In Image Processing

(ICIP), 2013 20th IEEE International Conference on,

pages 4205–4209.

Nigam, A. and Gupta, P. (2014a). Multimodal personal au-

thentication using iris and knuckleprint. In Huang,

D.-S., Bevilacqua, V., and Premaratne, P., editors, In-

telligent Computing Theory, volume 8588 of Lecture

Notes in Computer Science, pages 819–825. Springer

International Publishing.

Nigam, A. and Gupta, P. (2014b). Palmprint recognition

using geometrical and statistical constraints. In Babu,

B. V., Nagar, A., Deep, K., Pant, M., Bansal, J. C.,

Ray, K., and Gupta, U., editors, Proceedings of the

Second International Conference on Soft Computing

for Problem Solving (SocProS 2012), December 28-

30, 2012, volume 236 of Advances in Intelligent Sys-

tems and Computing, pages 1303–1315. Springer In-

dia.

Nigam, A. and Gupta, P. (2015a). Designing an accu-

rate hand biometric based authentication system fus-

ing ﬁnger knuckleprint and palmprint. Neurocomput-

ing, 151, Part 3:1120 – 1132.

Nigam, A. and Gupta, P. (2015b). Designing an accu-

rate hand biometric based authentication system fus-

ing ﬁnger knuckleprint and palmprint. Neurocomput-

ing, 151, Part 3:1120 – 1132.

Nigam, A., T., A., and Gupta, P. (2013). Iris classiﬁcation

based on its quality. In Huang, D.-S., Bevilacqua, V.,

Figueroa, J., and Premaratne, P., editors, Intelligent

Computing Theories, volume 7995 of Lecture Notes

in Computer Science, pages 443–452. Springer Berlin

Heidelberg.

Pizer, S. M., Amburn, E. P., Austin, J. D., Cromartie, R.,

Geselowitz, A., Greer, T., Romeny, B. H., Zimmer-

man, J. B., and Zuiderveld, K. (1987). Adaptive

histogram equalization and its variations. Computer

Vision, Graphics, and Image Processing, 39(3):355–

368.

PolyU (2010). The polyu knuckleprint database.

http://www4.comp.polyu.edu.hk/∼biometrics/.

Prasad, V. S. N. and Domke, J. (2007). Gabor ﬁlter visual-

ization.

Woodard, D. L. and Flynn, P. J. (2005). Finger surface as a

biometric identiﬁer. Computer Vision and Image Un-

derstanding, 100:357–384.

Xiong, M., Yang, W., and Sun, C. (2011). Finger-knuckle-

print recognition using lgbp. In International Sympo-

sium on Neural Networks, ISNN, pages 270–277.

Zhang, L., Zhaang, L., and Zhang, D. (2009a). Finger-

knuckle-print veriﬁcation based on band-limited

phase-only correlation. In International Conference

on Computer Analysis of Images and Patterns (CAIP),

pages 141–148.

Zhang, L. and Zhang, D. (2009). Finger-knuckle-print: A

new biometric identiﬁer. In International Conference

Image Processing (ICIP), pages 1981–1984.

Zhang, L., Zhang, L., and Zhang, D. (2009b). Finger-

knuckle-print veriﬁcation based on band-limited

phase-only correlation. In International Conference

on Computer Analysis of Images and Patterns, pages

141–148.

Zhang, L., Zhang, L., and Zhang, D. (2010a). Monogenic-

code: A novel fast feature coding algorithm with ap-

plications to ﬁnger-knuckle-print recognition. In In-

ternational Workshop on Emerging Techniques and

Challenges for Hand-Based Biometrics, pages 1 –4.

Zhang, L., Zhang, L., Zhang, D., and Guo, Z. (2012). Phase

congruency induced local features for ﬁnger-knuckle-

print recognition. Pattern Recognition, 45(7):2522–

2531.

Zhang, L., Zhang, L., Zhang, D., and Zhu, H. (2010b). On-

line ﬁnger-knuckle-print veriﬁcation for personal au-

thentication. Pattern Recognition, 43(7):2560–2571.

Zhang, L., Zhang, L., Zhang, D., and Zhu, H. (2010c). On-

line ﬁnger-knuckle-print veriﬁcation for personal au-

thentication. Pattern Recognition, 43(7):2560–2571.

Zhang, L., Zhang, L., Zhang, D., and Zhu, H. (2011a).

Ensemble of local and global information for ﬁnger-

knuckle-print recognition. Pattern Recognition,

44(9):1990 – 1998.

Zhang, L., Zhang, L., Zhang, D., and Zhu, H. (2011b).

Ensemble of local and global information for ﬁnger-

knuckle-print recognition. Pattern Recognition,

44(9):1990 – 1998.

Finger-Knuckle-Print ROI Extraction using Curvature Gabor Filter for Human Authentication

373