Content uploaded by Mouad M.H. Ali
Author content
All content in this area was uploaded by Mouad M.H. Ali on Jun 27, 2018
Content may be subject to copyright.
An Approach to Improve Palmprint Recognition
Accuracy by using Different Region of Interest
Methods with Local Binary Pattern Technique
Mouad M.H. Ali1*, A.T. Gaikwad2 and Pravin Yannawar1
1Department of CS&IT, Vision and Intelligent System Lab, Dr. Babasaheb Ambedkar Marathwada University,
Aurangabad − 431001, Maharashtra, India;
Mouad198080@gmail.com, pravinyannawarr@gmail.com,
2Institute of Management Studies and Information Technology, Aurangabad − 431001, Maharashtra, India;
drashoktgaikwad@gmail.com
Indian Journal of Science and Technology, Vol 11(22), DOI: 10.17485/ijst/2018/v11i22/122752, June 2018
ISSN (Print) : 0974-6846
ISSN (Online) : 0974-5645
Abstract
Objective: To extract the Region of Interest (ROI) of palmprint image by using appropriate methods and to improve the
accuracy of palmprint recognition system. Methods/Statistical Analysis: This piece of work is primarily addressing
the different mechanisms for extracting ROI area. The techniques like Competitive Hand Valley Detection (CHVD),
and Euclidean Distance (ED) were applied as the part of pre-processing, while the Feature Extraction mechanism LBP
was utilized to extract the texture feature from different type of ROIs of palmprint image. Findings: The experimental
results showed that CHVD with LBP gave best result with high accuracy reached to 96.10534% and Equal Error Rate
(EER) of 3.894661%, while in ED the best result showed accuracy reached to 88.23611% and EER of 11.76389%.
Application/Improvements: The study mainly concentrated on developing palmprint authentication system with
less EER and high accuracy.
Keywords: Analysis, Competitive Hand Valley Detection (CHVD), Euclidean Distance, Local Binary Pattern (LBP),
Palmprint Recognition
1. Introduction
Today, personal identification playsan active research and
applications in our daily life like banking, immigration,
and access control, ID card, passport office and
country borders etc. Recently, the biometric application
(identification or verification) has widely used and
majority of research studies have conducted in the field
of security system with different technologies (modalities
or traits) such as fingerprint, iris, palmprint, face, retina
etc1. Biometric is the automatic system which is used
to recognize the persons by their characterizations
(behavioral and physical)2. The main objective of any
biometric system is to achieve low cost, less error rate,
speed in performance and high accuracy.
There are two types of palmprint which called high
resolution and low resolution. Each type is suitable for
different applications. High resolution images are used
for forensicapplication3 while low resolution images are
used for access controls application.
Palmprint trait has a rich of information which
is useful for identification and verification propose.
Different types of feature can be extracted from high
and low resolution palmprint images. The features
from high resolution images include Minutiae feature,
Ridges features and Singular point feature4–5. Whereas,
the features from low resolution images include
Principal line, Texture feature, Palm-crease, Wrinkles
and Statistical features. All these features present on the
surface of palmprint images.
*Author for correspondence
An Approach to Improve Palmprint Recognition Accuracy by using Different Region of Interest Methods with Local Binary
Pattern Technique
Indian Journal of Science and Technology
Vol 11 (22) | June 2018 | www.indjst.org
2
The methods of palmprint can be classified into
different classes depending upon verification and
identification. The methods used in verification system
include Line based methods, subspace based methods
and statistical based methods. Line-based methods are
focused on edge detectors to extract the palm lines
from the palmprint images such as mention in6–14.
Matching of these lines can be made either directly
or indirectly by using different matching format.
Subspace based methods, PCA method, LDA method
and ICA method were studied in earlier works15–18
other methods likewavelets, Gabor, DCT and Kernels
were mentioned in19–22.
Statistical methods are either local or global. In
the case of Local the images should be transformed to
another domain and then divided into several small
regions23–27. The means, standard divisions and variances
of each region were calculated and stored as features.
Some of researchers used Gabor, wavelets and Fourier
transforms28–30. Another researcher used Local Binary
Pattern (LBP) histograms as features31. Whereas, some
of them used global statistical methods like Moments,
centers of gravity and density as mention in31–36.
The remaining sections of the paper are discussed the
methodology system which includes the preprocessing,
ROIs methods, feature extraction, matching and decision
that covered in section 2. The results and discussion
are given in details in section 3. Section 4 exhibits the
conclusion and future work.
2. Methodology of the System
In this section the methodology of the palmprint recogni-
tion system is discussed. The ROI from palmprint images
was extracted by two ways called as CHVD, andED. The
ROI was passed to feature extraction techniques to extract
the texture feature by using LBP technique, and then the
feature reduction technique was utilized to reduce the
dimensionality by PCA method. After that the features
were stored as data template which used for matching
purposes. This section was divided into different sub-
sections. Section 2.1 which covers the ROI methods in
details, section 2.2 which discusses LBP feature extraction
methods, section 2.3 which represents the matching pro-
cess and section 2.4 which shows the decision. Figure1
shows the block diagram of methodology of plamprint
recognition system.
2.1 ROI Extraction Methods
The ROI refers to a subset of all the data used in an obser-
vation. In palmprint ROI is the data of two- dimensional
and defines as rectangular area on the hand surface. It
is a small area which includes more information like
Minutiae, Principal line, and the key point. The key
points involve an end points or a branch points37 which
needed in the process of identification or verification.
There are many techniques utilized to extract the ROI
from palm image such as CHVD and ED. There are two
general steps performed before making ROI algorithm
Figure 1. Methodology system of palmprint recognition system.
Mouad M.H. Ali, A.T. Gaikwad and Pravin Yannawar
Indian Journal of Science and Technology 3
Vol 11 (22) | June 2018 | www.indjst.org
namely: image Binarization and boundary extraction
which include in the pre-processing stage as shown in
(Figure 2).
Image Binarization: It is the process of converting
gray scale images into black-and-white image based
on threshold value by multiplying a coefficient α with
the greatest possible values of the scale of gray used
(θ= α * 2n-1) where n is a fine gray scale used and θ is
the threshold value (T). The value is changed for each
pixel in image by Eq. (1) where x is the pixel values at
particular coordinates. 0 represents black and 1 repre-
sents white.
fx
x
x
()
=≥
0
1
;
;
<
θ
θ
(1)
Boundary Extraction: Boundary objects can be
generated from the black-and-white image by applying
erosion operator on the matrix structuring element. The
boundary of palmprint image can be obtained by Eq. (2):
Boundary
BW imageBWimage BW imag
eS
() ()
=− Θ (2)
Where θ is an operator to perform erosion on a black
and white image and S is the structuring element matrix
which can be represented as shown in Eq. (3).
S=
010
111
010
(3)
2.1.1 Competitive Hand Valley Detection (CHVD)
The CHVD algorithm was used to extract the ROI area
from palm image based on four reference points which
called the valley points that obtained by tracking the
boundary and seeking the coordinates of all the points.
The valley point is a pixel where most of its neighboring
pixels are in the object area of palmprint. Figure 3 repre-
sents a block diagram of extracting the ROI using CHVD
technique.
A CHVD algorithm checks whether a pixel included a
valley point by different cases as discusses below:
Case 1: Putting four points test with same distance
from pixel (current pixel) that will be checked whether
Figure 2. e main steps of palmprint pre-processing.
Figure 3. Block diagram of CHVD algorithm.
An Approach to Improve Palmprint Recognition Accuracy by using Different Region of Interest Methods with Local Binary
Pattern Technique
Indian Journal of Science and Technology
Vol 11 (22) | June 2018 | www.indjst.org
4
the pixel is a valley point. If one point is in the background
while other areas are in the area then the current pixel
palmprint be a candidate as a valley point and then do
check the current pixel to the next condition, if not then
the checks carried out for the next pixel,
Case 2: Test point increase to eight by the distance
from the current β + α pixel. Make checks if at least
one fruit or not more than four test points are in the
background area and the rest are in the area palmprint
the current pixel still be a candidate valley point and go
to the next check,
Case 3: Test point increased to 16 with a distance
β + α + µ pixel (current pixel). If at least one fruit or not
more than seven points are located on the background
area and the rest are in the area palmprint the current
pixel still be a candidate valley point and continued to
check the latest condition, and
Case 4: Pull a straight line from the current pixel to
the area of non-palmprint. If the line does not intersect
with the current point palmprint area is considered as the
valley point.
CHVD algorithm is based on the assumption that no
person flexing his fingers over 120o 38. The main problem
in the CHVD algorithm is certain conditions can produce
more than one candidate valley point. The solution is to
select a point depend on the y axis value of the greatest
or is at the bottom of the image of the four valley point
{P1, P2, P3, P4} two points taken as a reference point that
is P1 and P2 as shown in Figure 4. ROI generation differs
in terms of size and orientation. Then, the next step is to
use the amount of the angle rotates ROI obtained from
two reference points and changing all the ROI into a stan-
dard size using the Bicubic Interpolation. Figure5 shows
our developing of the Graphical User Interface (GUI) of
automatic system for extracting ROI by using CHVD
method.
2.1.2 Euclidean Distance
The idea of taking ROI using Euclidean Distance method
is the same as CHVD technique. However, they differ by
the way of getting the reference point. Figure 6 shows
the block diagram of the ROI retrieval process by using
Euclidean Distance.
Figure 4. Making ROI process by CHVD.
Figure 5. GUI of automatic ROI system by CHVD
algorithm.
Figure 6. Block diagram of Euclidean distance.
Mouad M.H. Ali, A.T. Gaikwad and Pravin Yannawar
Indian Journal of Science and Technology 5
Vol 11 (22) | June 2018 | www.indjst.org
The way to determine the centroid point of an object
in a black-and-white image was to find the mean values of
x and y of the overall coordinates point which is the loca-
tion of the pixel that represents the object. The centroid
coordinate point can be searched using Eq. (4).
xNxy Ny
centroid
i
N
icentroid
i
N
i
==
==
∑∑
11
11
and (4)
Where xi and yi are the values of x and y at p (x, y) = 1.
The next step was to calculate the distance between
each point on the boundary with its centroid point using
the Eq. (5).
Dist xy xx yy
centroid centroid
,()
()
()
=+−−
22
(5)
All the distance values resulting from Euclidean dis-
tance were calculated and plotted as shown in the Figure
7a.
The peak point shows the pixel location of the fingertips
while the valley point indicates the pixel location of the
valley point seen from its distance to the centroid point.
To obtain the coordinates of the valley point, first find the
first derivative of the Euclidean function and then mark
the dots passing the zero on the y-axis (zero-crossing
points). Figure 7b shows the first derivative value of the
Euclidean function.
Both the vertex and valley points passed through the
value 0 on the y-axis. However, in the graph above a lot
of zero-crossing caused by the erratic contours of the
palmprint image39 so that the valley point location cannot
be determined yet. This can be overcome by smoothing
the graph by removing high frequency components and
maintaining low frequency components. Figure 7c shows
the results of the smoothing.
Valley point can be identified from the point that
passed the value 0 on the y-axis and changed the sign from
negative to positive. The reference point used is only the
valley point between the index finger and the middle finger
(V1) and between the ring finger and the little finger (V3) as
shown in Figure 8.
Figure 7. Extract ROI Method by ED: (a) Euclidean distance of each pixel within the boundary
with the centroid, (b) First derivative value, and (c) smoothing value of First derivative graph.
Figure 8. ROI detection of a palm image by ED.
An Approach to Improve Palmprint Recognition Accuracy by using Different Region of Interest Methods with Local Binary
Pattern Technique
Indian Journal of Science and Technology
Vol 11 (22) | June 2018 | www.indjst.org
6
Then point V1 and V3 were used to find the angle of
rotation so that the ROI is the same for the hand with
different orientations. The final step was to draw a
straight line from point V1 to point V3 to form the ROI
area. Figure 9 shows our developing system of extracting
the ROI by Euclidean Distance method.
2.2 Feature Extraction using LBP
LBP operator considered as a texture feature and it is
used for shape extraction of gray scale image. It refers
to a binary code for an image-pixel and it provides us
by information regarding the local neighborhood of
that pixel. The principal LBP operator was given by40.
The idea behind the LBP operator is to search a center
pixel value of an image and take this value as threshold
and check as below case:
LBPBinaryCode
if neigborhood
pixelThreshold
Otherwi
__
;
;
=>=
1
0
sse
If neighbor pixel has higher value or equal to center
pixel value, the pixel takes the value 1, otherwise it takes
the value zero. The example of LBP and how to calculate
a binary code is represented in Figure 10.
After that the LBP was extended to utilize neighbor-
hoods of various sizes. In such case a circle is produced
having radius R from the center pixel. P sampling points
on the circle edge are determined and compared with cen-
ter pixel value to find the values of all sampling points in
neighborhoods for any radius and any number of pixels.
Figure 11 demonstrates three neighbor sets for various val-
ues of P and R.
Figure 10. Stages of LBP calculation process.
Figure 11. Circular (8,1), (8,2) and (16,2) neighborhoods.
Figure 9. GUI of Automatic ROI system by using Euclidean
Distance Algorithm.
Mouad M.H. Ali, A.T. Gaikwad and Pravin Yannawar
Indian Journal of Science and Technology 7
Vol 11 (22) | June 2018 | www.indjst.org
In the case of our work the LBP feature extraction was
used to calculate the LBP for every pixel in the image.
This occurred by divided the palmprint image into blocks
or regions. Here the image was divided into (8 × 8) block
size. The feature vector of the image can be generated by
combining all histogram of each block. Figure 12 shows
the process of our LBP technique.
Algorithm 1 describes the LBP step by step.
Algorithm 1. Feature Extraction using LBP
1. Input: Palmprint images ROI of CHVD and ED.
2. Output: CHVD+LBP Features, ED+LBP features
3. Begin
4. For each sample of CHVD and ED images do
5. Divided palmprint images to(8 × 8) overlap blocks
6. For each pixel in the blocks do
7. Compare pixel to all 8 neighbors
8. IF Center pixel > neighbors then
9. Replace the neighbors to 1
10. Else
11. Replace the neighbors to 0
12. End
13. Generate the binary code from all the neighbors and
convert it to decimal
14. Apply histogram for all the cell (in which their
neighbors grater of smaller to center pixel)
15. FV LBPCHVDROI
CHVD LBPhisti
i
N
+
=
=
()
() _
1
∪
16. FV LBPEDROI
ED LBPhisti
i
N
+
=
=()
(_ )
1
∪
17. Store the FVCHVD LBP+and FVED LBP+as feature vector.
18. End
19. End
2.3 Matching Process
The matching was done by compared input image (Query
or Test) with the template which stored in the database
that taken at the time of enrollment and compute the
degree of similarity or dissimilarity from two templates.
To achieve the score we used Euclidean distance measure
to compute the similarity or the difference between the
templates, The Euclidean distance can be calculated by
using Eq. (6).
dT
estTempTestV Template V
ij
ij
ij
N
()(_
_)
,
&
=−
=
∑2
1
(6)
Where, N is the number of feature in
Test Vi
_
and
Template Vj
_
.
Algorithm 2 describes the Matching step by step.
Algorithm 2. Matching
1. Input: Palmprint feature vector
2. Output: Score matrix, reshold values.
3. Begin:
4. For each test feature of each Subjectdo
5. Test_FVi
6. For each template feature of each Subjectdo
7. Template_FVj
dist Test Temp Test FV Template FV
ij
ij
ij
n
(, )_ _()
&
=−
=
∑2
1
8. Score_matrix= d (Tes t i, Te mpj)
9. End
10. T0=Score_matrix;/* total threshold values of system
11. minta = min(min(T0)); /* minimum score
Figure 12. LBP of palmprint image with histogram of each block.
An Approach to Improve Palmprint Recognition Accuracy by using Different Region of Interest Methods with Local Binary
Pattern Technique
Indian Journal of Science and Technology
Vol 11 (22) | June 2018 | www.indjst.org
8
12. maxta = max(max(T0)); /* maximum score
13. β = 100; /* size of optimal threshold
values
14. Δ = (maxta - minta) / β;
15. const = 1 : 1 : β; /* threshold vector
16. End
17. Store score_matrix in database as math le for
discussion purposed.
18. End
2.4 Decision
In the final steps of the study the decision was taken
either “Accepted “or “Rejected “with the help of threshold
value (T). In the case of “Accepted”, the distance (score S)
should be(S ≤ T)otherwise it means “Rejected”.
3. Results and Discussion
This section divided into different subsections. In section
3.1 dataset details are covered. The evaluation matrix is
addressing in section 3.2, the results of extracting ROI
by CHVD with LBP feature extraction is discussed in
section3.3. The section 3.4 addresses the results of ROI by
ED and feature extraction techniques by using also LBP.
The experimental applied on the laptop Dell, Intel core
i3, CPU 2.20 GHz with RAM 8.00GB on 64-bit operating
system (windows 7) and by MATLAB software 2013a.
3.1 Database
The experiments were evaluated on Chinese Academy
of Sciences’ Institute of Automation (CASIA)41,42 Multi-
Spectral Database v1.0 palmprint database which has 8
bits gray level (JPEG file), image size (768 × 576). This
database contains 7200 images for 100 subjects for both
left and right hands. For evaluate the algorithm the 100
subjects were selected for left hand with ID 001–100 and
each subjects has 6 samples each labeled 01–06 and the
total dataset containing 1200 images. The experimental
applied on the laptop Dell, Intel core i3, CPU 2.20 GHz
with RAM 8.00GB on 64-bit operating system (win-
dows 7). Figure 13 shows some samples taken from
CASIA dataset and their ROI part.
3.2 Evaluation Matrix
To evaluate any biometric system related to specific
application there are different parameters namely False
Accepted Rate (FAR), False Rejected Rate (FRR) and
Equal Error Rate (EER). These parameters should be
achieved the lowest values to get the best performance of
the system. The FAR is the ratio of imposter score exceed-
ing the threshold values divided by all the imposter score
generated by the system and calculated by Eq. (7).
FAR
=×
Impostor Score exceeding thershold
All Impostor Score 1000 (7)
The FRR is the ratio of genuine score falling below the
threshold value divided by all the genuine score generated
by the system and calculated by Eq. (8).
FRR=
Genuine Scores falling below thershold
All Genuine Scorees
×100 (8)
The EER can be calculated according to the Eq. (9)
EER
FARFRR
=
+
2
(9)
In addition to the above parameters there is Genuine
Accept Rate (GAR) which shows the relation between
FAR and FRR with the help of threshold values. Also the
Receiver Operating Characteristic (ROC) curves shows
the FAR values which are changed related to GAR values
and it shows the performance of the system. The GAR
value is calculated by Eq. (10).
GARFRR=−1
(10)
Figure 13. Samples taken from CASIA dataset with extract ROI.
Mouad M.H. Ali, A.T. Gaikwad and Pravin Yannawar
Indian Journal of Science and Technology 9
Vol 11 (22) | June 2018 | www.indjst.org
3.3 Results of CHVD ROI with LBP
The performance of the system was evaluated by CHVD
method with LBP feature extraction technique. The size
of ROI was (155 × 155) which applied on dataset size of
100 users, each user has 6 samples. The threshold values
were generated from the score matrix at the matching
stages. There are 100 threshold values ranged from 0 to 1.
The system can achieved the best performance with
minimum EER and maximum GAR which determined
with help of threshold values. The system achieved dif-
ferent results on different threshold values e.g. on T = 0
both EER and GAR of the system were 50% that means
higher EER and minimum GAR of the system. The sys-
tem achieved the best performance with minimum EER
of 3.894661% and maximum GAR of 96.10534% at the
threshold value of0.6. Tabl e 1 shows the results of the
system by different T values with FAR, FRR, EER and
GAR. Figure 14a depicts the relation between FAR and
FRR with the help of threshold values and Figure 14b
shows the ROC of the system by GAR Vs. FAR on
different threshold values.
3.4 Results of ED with LBP
In the second experimental, the ED method was used
to extract the ROI of (131 × 131) size then passed to
LBP feature extraction method which evaluated by the
same dataset applied in previous experiment. The sys-
tem was achieved different results on different T values
reached to 100and ranged from 0 to 1 which increased
by 0.01 in each iteration and these threshold values
were used to check the performance of the system. The
efficiency result was achieved by minimum EER of
Table 1. Performance of the system based on CHVD
with LBP Feature
CHVD + LBP
T FAR (%) FRR (%) EER (%) GAR (%)
0 0 100 50 50
0.1 0.035714 86.18182 43.10877 56.89123
0.2 0.071429 62.44444 31.25794 68.74206
0.3 0.964286 36.26263 18.61346 81.38654
0.4 2.214286 17.83838 10.02633 89.97367
0.5 3.464286 8 5.732143 94.26786
0.6 5.142857 2.646465 3.894661 96.10534
0.7 9.428571 0.707071 5.067821 94.93218
0.8 13.42857 0.242424 6.835498 93.1645
0.9 19.39286 0.10101 9.746934 90.25307
126.60714 0.040404 13.32377 86.67623
Figure 14. Performance of CHVD+LBP System: (a) Relation of FAR vs. FRR based on threshold value, and (b) ROC
curve of system based on FAR vs. GAR.
(a) (b)
An Approach to Improve Palmprint Recognition Accuracy by using Different Region of Interest Methods with Local Binary
Pattern Technique
Indian Journal of Science and Technology
Vol 11 (22) | June 2018 | www.indjst.org
10
Figure 15. Performance of ED+LBP: (a) Relation of FAR vs. FRR based on threshold value, and (b) ROC curve of FAR
vs. GAR for.
(a) (b)
Figure 16. Comparison of GAR of CHVD and ED algorithm with LBP feature extraction: (a) EER curve, and
(b) GAR Vs. FAR.
(a) (b)
11.76389% and Maximum GAR reached to 88.23611%
which showed the best results compared with the previ-
ous technique. Tab l e2 shows the results of the system
by different T values with FAR, FRR, EER and GAR and
(Figure 15a) depicts the relation between FAR and FRR
with help of threshold values and (Figure 15b) shows
the ROC of the system by GARVs, FAR on different
threshold values.
Finally, the comparison between both ROI methods
shows that ED with LBP feature technique gave the least
result with higher EER reached to 11.76389% and GAR
of 88.23611% while CHVD with LBP achieved the best
result with minimum EER of 3.894661% and maximum
GAR of 96.10534%. Thecomparison between both meth-
ods based on EER and GAR is shown in (Figure16a) and
(Figure 16b) respectively.
Table 2. Performance of the system based on ED
with LBP feature
ED + LBP
T FAR (%) FRR (%) EER (%) GAR (%)
0 0 100 50 50
0.1 0.178571 95.0101 47.59434 52.40566
0.2 9.25 18.0202 13.6351 86.3649
0.22 11.75 11.77778 11.76389 88.23611
0.3 26 2.383838 14.19192 85.80808
0.4 46.32143 1.111111 23.71627 76.28373
0.5 67.82143 0.666667 34.24405 65.75595
0.6 83.85714 0.363636 42.11039 57.88961
0.7 94.35714 0.20202 47.27958 52.72042
0.8 99 0.060606 49.5303 50.4697
0.9 99.96429 0 49.98214 50.01786
1 100 0 50 50
Mouad M.H. Ali, A.T. Gaikwad and Pravin Yannawar
Indian Journal of Science and Technology 11
Vol 11 (22) | June 2018 | www.indjst.org
4. Conclusion
This research work evaluated the performance of two
ROI techniques namely CHVD and ED on CASIA Multi-
Spectral Database v1.0. The extraction of ROI is very
important step in palmprint recognition system. It helps
in extraction of features in terms of real part which con-
tains rich information required for an authentication
system. The research work evaluated ROI extracted by
using CHVD and ED techniques, build feature matrix
using LBP technique and PCA method which used for
feature reduction. The LBP was applied for both ROIs and
it was concluded from the experimental work that CHVD
method achieved better performance with GAR reached
to 96.10534% with EER equal to 3.894661%, while the
EDROI method applied to LBP and PCA gave GAR
reached to 88.23611% with EER equal to 11.76389%. This
work may be extended to combine LBP with LPP features
with neural network to achieve the best performance of
the system.
5. References
1. Jain A, Bolle R, Pankanti R. Biometrics: Personal Identication
in Networked Society. Norwell, Massachusetts: Kluwer
Academic Publishers; 1999. p. 1−411. Crossref, Crossref.
2. Anil KJ, Flynn P, Ross AA. Handbook of biometrics.
Springer, USA; 2007. p. 1−566.
3. NEC Automated Palmprint Identication System. Date
accessed: 01/012000. https://www.policemag.com/channel/
technology/products/2000/01/automated-palmprint-iden-
tication-system.aspx.
4. Duta N, Jain AK, Mardia KV. Matching of palmprints,
Pattern Recognition Letters. 2002; 23(4):477–85. Crossref.
5. Shu W, Zhang D. Automated personal identication by
palmprint, Optical Engineering. 1998; 38(8):2359–62.
Crossref.
6. Li F, Leung MKH, Yu X. Palmprint identication using
Hausdor distance. In: Proceedings of IEEE International
Workshop on Biomedical Circuits and Systems; 2004.
p. S3/3–S5-8. PMid:15336835.
7. Wu X, Wang K, Zhang D. Line feature extraction and
matching in palmprint. In: Proceeding of the Second
International Conference on Image and Graphics; 2002.
p. 583–90. Crossref.
8. Wu X, Wang K, Zhang D. Fuzzy direction element energy
feature (FDEEF) based palmprint identication, Object
Recognition Supported by User Interaction for Service
Robots. 2002; 1:95–98.
9. Boles WW, Chu SYT. Personal identication using images
of the human palms. In: Proceedings of IEEE Region 10
Annual Conference, Speech and Image Technologies for
Computing and Telecommunications; 1997. 1, p. 295–98.
Crossref.
10. Rafael Diaz M, Travieso CM, Alonso JB, Ferrer MA.
Biometric system based in the feature of hand palm.
In: Proceedings of 38th Annual International Carnahan
Conference on Security Technology; 2004. p. 136–39.
11. Kung SY, Lin SH, Fang M. A neural network approach to
face/palm recognition. In: Proceedings of IEEE Workshop
Neural Networks for Signal Processing; 1995. p. 323–32.
Crossref.
12. Rodrigues BAR, Silva JDL. Biometric identication by der-
matoglyphics. In: Proceedings of International Conference
on Image Processing; 1996. 1, p. 319–22.
13. Wu W, Wang K, Zhang D. A novel approach of palm-
line extraction. In: Proceeding of the ird International
Conference on Image and Graphics; 2004. p. 230–33.
PMid:15193533.
14. Ali MHM, Yannawar P, Gaikwad AT. Study of edge detection
methods based on palmprint lines. In: Proceedings of
International Conference on Electrical, Electronics, and
Optimization Techniques (ICEEOT). Chennai, India; 2016.
p. 1344−50. Crossref.
15. Connie T, Jin ATB, Ong MGK, Ling DNC. An auto-
mated palmprint recognition system, Image and Vision
Computing. 2005 May; 23(5):501–15. Crossref.
16. Wu X, Zhang D, Wang K. Fisherpalms based palmprint
recognition, Pattern Recognition Letters. 2003 Nov;
24(15):2829−38. Crossref.
17. Lu G, Zhang D, Wang K. Palmprint recognition using
Eigenpalms features, Pattern Recognition Letters. 2003 Jun;
24(9-10):1463−67. Crossref.
18. Jing XY, Zhang D. A face and palmprint recognition
approach based on discriminant DCT feature extraction,
IEEE Transactions on Systems, Man, and Cybernetics, Part
B (Cybernetics). 2004 Dec; 34(6):2405−15. Crossref.
19. Li Q, Qiu Z, Sun D. Feature-level fusion of hand bio-
metrics for personal verication based on Kernel PCA.
International Conference on Biometrics. Springer, Berlin;
2006. p. 744−50.
20. Lu GM, Wang KQ, Zhang D. Wavelet based indepen-
dent component analysis for palmprint identication.
In: Proceedings of International Conference on Machine
Learning and Cybernetics; 2004 Aug. 6, p. 3547–50. Crossref.
21. Chu R, Lei Z, Han Y, Li SZ. Learning Gabor magnitude
features for palmprint recognition. In Asian Conference
on Computer Vision ACCV. Springer, Berlin, Heidelberg;
2007. p. 22−31. Crossref.
An Approach to Improve Palmprint Recognition Accuracy by using Different Region of Interest Methods with Local Binary
Pattern Technique
Indian Journal of Science and Technology
Vol 11 (22) | June 2018 | www.indjst.org
12
22. Ekinci M, Aykut M. Palmprint recognition by apply-
ing wavelet sub and representation and kernel PCA.
In International Workshop on Machine Learning and
Data Mining in Pattern Recognition. Springer, Berlin,
Heidelberg; 2007 Jul. p. 628−42. Crossref, Crossref.
23. Han CC, Cheng HL, Lin CL, Fan KC. Personal authentica-
tion using palm-print features, Pattern Recognition. 2003;
36(2):371–81. Crossref.
24. You J, Kong WK, Zhang D, Cheung KH. On hierarchical
palmprint coding with multiple features for personal iden-
tication in large databases, IEEE Transactions on Circuits
and Systems for Video Technology. 2004; 14(2):234–43.
Crossref.
25. Lu G, Wang K, Zhang D. Wavelet based feature extrac-
tion for palmprint identication. In: Proceeding of Second
International Conference on Image and Graphics; 2002.
p. 780−85. Crossref.
26. Wu X, Wang K. Zhang D. Palmprint recognition using
directional energy feature. In: Proceedings of International
Conference on Pattern Recognition; 2004. 4, p. 475–78.
27. Mouad MHA, Gaikwad AT. Multimodal Biometrics
Enhancement Recognition System based on Fusion of
Fingerprint and PalmPrint: A Review, Global Journal of
Computer Science and Technology. 2016 Sep; 16(2):13−26.
28. Poon C, Wong DCM, Shen HC. Personal identication
and verication: fusion of palmprint representations. In:
Proceedings of International Conference on Biometric
Authentication, Springer, Berlin, and Heidelberg; 2004.
p. 782–88. Crossref.
29. Kumar A, Shen HC. Palmprint identication using
PalmCodes. In: Proceedings of 3rd International Conference
on Image and Graphics; 2004. p. 258–61. Crossref.
30. Mouad MHA, Yannawar PL, Gaikwad AT. Multi-Algorithm
of Palmprint Recognition System Based on Fusion of Local
Binary Pattern and Two-Dimensional Locality Preserving
Projection, Procedia Computer Science. 2017 Dec;
115:482−92. Crossref.
31. Wang X, Gong H, Zhang H, Li B, Zhuang Z. Palmprint
identication using boosting local binary pattern. In:
Proceedings of International Conference on Pattern
Recognition; 2006 Aug. 3, p. 503–06.
32. Pang YH, Connie T, Jin A, Ling D. Palmprint authentica-
tion with Zernike moment invariants. Proceedings of the
3rd IEEE International Symposium on Signal Processing
and Information Technology; 2003 Dec. p. 199–202.
33. Zhang L, Zhang D. Characterization of palmprints by
wavelet signatures via directional context modeling, IEEE
Transactions on Systems, Man and Cybernetics. 2004 Jun;
34(3):1335–47. Crossref.
34. Dai Q, Bi N, Huang D, Zhang D, Li F. M-band wavelets
applications to palmprint recognition based on texture
features. In: Proceedings Conference on Image Processing;
2004 Oct. 2, p. 893–96.
35. Li Y, Wang K, Zhang D. Palmprint recognition based on
translation invariant Zernike moments and modular neural
network. International Symposium on Neural Networks;
2005 May. p. 177–82. Crossref.
36. Noh JS, Rhee KH. Palmprint identication algorithm
using Hu invariant moments and Otsu binarization.
In: Proceeding of Fourth Annual ACIS International
Conference on Computer and Information Science; 2005.
p. 94–99.
37. Zhang DD. Palmprint authentication. Springer Science and
Business Media; 2004. p. 1−241.
38. Wirayuda TAB, Habbi AA, Didik HK, Retno ND.
Real-time hand-tracking on video image based on
palm geometry. In Information and Communication
Technology (ICoICT), International Conference; 2013
Mar. p. 241−46. Crossref.
39. Xin P, Qiu-Qi R. Palmprint recognition using Gabor
feature-based (2D) 2 PCA, Neurocomputing. 2008 Aug;
71(13):3032−36.
40. Timo O, Pietikäinen M, Harwood D. A comparative study
of texture measures with classication based on featured
distributions, Pattern recognition. 1996 Jan; 29(1):51−59.
Crossref.
41. Biometrics Ideal Test. Fingerprint Image Database Version
5.0. Date accessed: 2014. http://biometrics.idealtest.org/.
42. Hao Y, Sun Z, Tan T, Ren C. Multi-spectral palm image
fusion for accurate contact-free palmprint recognition.
Proceedings of IEEE International Conference on Image
Processing; 2008 Oct. p. 281−84.