Introduction to IEEE STANDARDS and its different types.pptx
Iaetsd latent fingerprint recognition and matching
1. LATENT FINGERPRINT RECOGNITION AND MATCHING
USING STATISTICAL TEXTURE ANALYSIS
M.CHARAN KUMAR1
, K.PHALGUNA RAO 2
,
¹ M.Tech 2nd year, Dept. of CSE, ASIT, Gudur, India
² Professor, Dept. of CSE, ASIT, Gudur, India
1
cherry.abu@gmail.com; 2
kprao21@gmail.com;
Abstract: Latent’s are the partial fingerprints that are
usually smudgy, with small area and containing large
distortion. Due to this characteristic, latent’s have a
significantly smaller number of minutiae points
compared to full (rolled or plain) fingerprints. The
small number of minutiae and noise of latents make
it extremely difficult to automatically match latent’s
their mated full prints that are stored in law
enforcement databases, although a number of
methods used for fingerprint recognition to extract
accurate results but is not up to level. The proposed
fingerprint recognition and matching using statistical
analysis gives efficient scheme of fingerprint
recognition for biometric identification of
individuals. Three statistical features are extracted to
represent in mathematical model. They are (1) an
entropy coefficient, for intensity histogram of the
image, (2) a correlation coefficient, for operation
between the original and filter image by using wiener
filter, and (3) an energy coefficient, obtaining image
in 5-level wavelet decomposition obtained after 5th
level decomposition. The approach can be easily used
to provide accurate recognition results.
Index Terms: - fingerprint recognition, entropy,
correlation, wavelet energy.
1. INTRODUCTION
SIGNIFICANT improvements in fingerprint
recognition have been achieved in terms of
algorithms, but there are still many challenging tasks.
One of them is matching of nonlinear distorted
finger-prints. According to Fingerprint Verification
Competition 2004 (FVC2004), they are particularly
insisted on: distortion, dry, and wet fingerprints.
Distortion of fingerprints seriously affects the
accuracy of matching. There are two main reasons
contributed to the fingerprint distortion. First, the
acquisition of a fingerprint is a three-
dimensional/two-dimensional warping process. The
fingerprint captured with different contact centers
usually results in different warping mode. Second,
distortion will be introduced to fingerprint by the no
orthogonal pressure people exert on the sensor. How
to cope with these nonlinear distortions in the
matching process is a challenging task. Several
fingerprint matching approaches have been proposed
in the literature. These include methods based on
point pattern matching, transform features and
structural matching. Many fingerprint recognition
algorithms are based on minutiae matching since it is
widely believed that the minutiae are most
discriminating and reliable features. Rather al.
addressed a method based on point pattern matching.
The generalized Hough transform (GHT) is used to
recover the pose transformation between two
impressions. Jain et al. proposed a novel later bank
based fingerprint feature representation method.
Jingled. Addressed a method which relies on a
similarity measure defined between local structural
features, to align the two pat- terns and calculate a
matching score between two minutiae lists. Fantail.
Applied a set of geometric masks to record part of the
178
INTERNATIONAL CONFERENCE ON CURRENT INNOVATIONS IN ENGINEERING AND TECHNOLOGY
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT
ISBN: 378 - 26 - 138420 - 5
www.iaetsd.in
2. rich information of the ridge structure. What et al.
Addressed a method using groups of minutiae to
define local structural features. The matching is
performed based on the pairs of corresponding
structural features that are identified between two
fingerprint impressions. However, these methods do
not solve the problem of nonlinear distortions.
Recently, some algorithms have been presented to
deal with the nonlinear distortion in fingerprints
explicitly in order to improve the matching
performance. Proposed a method to measure the
forces and torques on the scanner directly. This
prevents capture with the aid of specialized hardware
when excessive force is applied to the scanner
Doraietal. Proposed a method to detect and estimate
distortion occurring in fingerprint videos, but those
two mentioned methods do not work with the
collected fingerprint images. Mao and Maltonietal.
Proposed a plastic distortion model to cope with the
nonlinear deformations characterizing finger- print
images taken from on-line acquisition sensors. This
model helps to understand the distortion process.
However, it is hard to automatically and reliably
estimate the parameter due to the insufficiency and
uncertainty of the information. Doggie Leeetal.
Addressed a minutiae-based fingerprints matching
algorithm using distance normalization and local
alignment to deal with the problem of the nonlinear
distortion. However, rich information of the
ridge/valley structure is not used, and the matching
performance is moderate.
However, in reality, approximately 10% [20] of
acquired fingerprint images are of poor quality due to
variations in impression conditions, ridge
configuration, skin conditions, acquisition devices,
and non-cooperative attitude of subjects, etc. The
ridge structures in poor-quality fingerprint images are
not always well defined and, therefore, cannot be
correctly detected. A significant number of spurious
minutiae may be created as a result. In order to
ensure that the performance of the minutiae
extraction algorithm will be robust with respect to the
quality of input fingerprint images, an enhancement
algorithm which can improve the clarity of the ridge
structures is necessary. However, for poor fingerprint
image, some spurious minutiae may still exist after
fingerprint enhancement and post processing. It is
necessary to propose a method to deal with the
spurious minutiae.
Fig. 1. Feature set of a live-scan fingerprint image. (a)
Original fingerprint image. (b) Thinned ridge image with
minutiae and sample points of (a).
A method to judge whether an extracted minutia is a
true one has been proposed in this paper. According
to our work, the distance between true minutiae is
generally greater than threshold (three). While near
the spurious minutiae, there are usually other
spurious minutiae. On the other hand, spurious
minutiae are usually detected at the border of
fingerprint image. Examples of spurious minutiae in
poor quality fingerprint images are shown in Fig. 2.
179
INTERNATIONAL CONFERENCE ON CURRENT INNOVATIONS IN ENGINEERING AND TECHNOLOGY
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT
ISBN: 378 - 26 - 138420 - 5
www.iaetsd.in
3. Fig. 2. Examples of spurious minutiae in poor quality
fingerprint images. The images have been cropped and
scaled for view. (a) Original image. (b) Enhanced image of
(a), (c) original image, (d) enhanced image of (c). Many
spurious minutiae were detected in the process of minutiae
extraction. Near the spurious minutiae, there are usually
other spurious minutiae as indicated in ellipses, and
spurious minutiae are usually detected at the border of
fingerprint image as shown in rectangles.
2. EXISTING SYSTEM
A. Fingerprint Recognition
The existing algorithm uses a robust alignment
algorithm (descriptor-based Hough transform) to
align fingerprints and measures similarity between
fingerprints by considering both minutiae and
orientation field information. To be consistent with
the common practice in latent matching (i.e., only
minutiae are marked by latent examiners), the
orientation field is reconstructed from minutiae.
Since the proposed algorithm relies only on manually
marked minutiae, it can be easily used in law
enforcement applications. Experimental results on
two different latent databases show that the proposed
algorithm outperforms two well optimized
commercial fingerprint matchers.
Fingerprint recognition (also known as
Dactyloscopy) is the process of comparing known
fingerprint against another or template fingerprint to
determine if the impressions are from the same finger
or not. It includes two sub-domains: one is fingerprint
verification and the other is finger print
identification. Verification specify an individual
fingerprint by comparing only one fingerprint
template stored in the database, while identification
specify comparing all the fingerprints stored in the
database. Verification is one to one matching and
identification is one to N (number of fingerprint
templates available in database) matching.
Verification is a fast process as compared to
identification.
Fig.2. Fingerprint Recognition System
Fig.2 shows the basic fingerprint recognition system.
First of all we take a fingerprint image. After taking
an input image we can apply fingerprint segmentation
technique. Segmentation is separation of the input
data into foreground (object of interest) and
background (irrelevant information). Before
extracting the feature of a fingerprint it is important
to separate the fingerprint regions (presence of
ridges) from the background. This is very useful for
recovering false feature extraction. In some cases, a
correct segmentation is very difficult, especially in
poor quality fingerprint image or noisy images.
Orientation field plays an important role in
fingerprint recognition system. Orientation field
180
INTERNATIONAL CONFERENCE ON CURRENT INNOVATIONS IN ENGINEERING AND TECHNOLOGY
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT
ISBN: 378 - 26 - 138420 - 5
www.iaetsd.in
4. consist of four major steps (1) pre processing
fingerprint image (2) determining the primary ridges
of fingerprint block (3) estimating block direction by
projective distance variance of such a ridge (4)
correcting the estimated orientation field. Image
enhancement is use to improve significantly the
image quality by applying some image enhancement
technique. The main purpose of such procedure is to
enhance the image by improving the clarity of ridge
structure or increasing the consistency of the fridge
orientation. Fingerprint classification is used to check
the fingerprint pattern type. After classification of
fingerprint. We can apply fingerprint ridge thinning
which is also called block filtering; it is used to
reduce the thickness of all ridges lines to a single
pixel width. Thinning does not change the location
and orientation of minutiae points compared to
original fingerprint which ensures accurate estimation
of minutiae points. Then we can extract minutiae
points and generate data matrix. Finally we can use
minutiae matching to compare the input fingerprint
data with the template data and give the result.
B. Fingerprint Matching Techniques
There are many Fingerprint Matching Techniques.
Most widely used matching techniques are these:
• Correlation-based matching:
In correlation based matching the two fingerprint
images are matched through corresponding pixels
which is computed for different alignments and
rotations. The main disadvantage of correlation based
matching is its computational complexity.
• Minutiae-based matching:
This is the most popular and widely used technique,
for fingerprint comparison. In minutiae-based
techniques first of all we find minutiae points on
which we have to do mapping. However, there are
some difficulties when using this approach. It is
difficult to identify the minutiae points accurately
when the fingerprint is of low quality.
• Pattern-based (or image-based) matching:
Pattern based technique compare the basic fingerprint
patterns (arch, whorl, and loop) between a previously
stored template and a candidate fingerprint. This
requires that the images be aligned in the same
orientation. In a pattern-based algorithm, the template
contains the type, size, and orientation of patterns
within the aligned fingerprint image.
The candidate fingerprint image is graphically
compared with the template to determine the degree
to which they match.
3. PROPOSED SYSTEM
A. entropy
Image entropy is an extent which is used to explain
the business of an image, i.e. the amount of
information which must be implicit for by a
compression algorithm. Low entropy Images, such as
those include a lot of black sky, have very little
difference and large runs of Pixels with the same or
parallel DN values. An Image that is entirely flat will
have entropy of Zero. Therefore, they can be
compressed to a relatively small size. On the other
hand, high
Entropy images such as an image of heavily formed
areas on the moon have a great deal of
Thing from one pixel to the next and accordingly
cannot be compressed as much as low entropy
images. Image entropy as used in this paper is
calculated with the same formula used by the Galileo
Imaging Team Entropy :
181
INTERNATIONAL CONFERENCE ON CURRENT INNOVATIONS IN ENGINEERING AND TECHNOLOGY
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT
ISBN: 378 - 26 - 138420 - 5
www.iaetsd.in
5. In the above expression, P is the probability
That the difference between two adjacent pixels
Is equal to i, and Log2i is the base 2 algorithm.
Entropy successfully bounds the performance of
The strongest lossless compression feasible, which
can be realized in theory by using the
Distinctive set or in perform using Huffman.
B. Correlation
Digital Image connection (correlation) and Tracking
(DIC/DDIT) is an optical method that employs
tracking & image check techniques for accurate 2-D
and 3-D measurements of change in images. This is
often used to measure deformation (engineering),
displacement, strain, and visual flow, but it is widely
applied in many areas of science and engineering
calculations. The image is first subjected to a 2-D
wiener filter using a 3*3 mask. By using wiener filter
to remove the redundant noise or un-wanted pixels.
Digital image correlation (DIC) techniques have been
increasing in status, especially in micro- and neon-
scale mechanical testing applications due to its
relative ease of implementation and use. Advances in
computer technology and digital cameras have been
the enabling technologies for this method and while
white-light optics has been the leading approach, DIC
can be and has been extensive to almost any imaging
technology.
The cross correlation coefficient is defined by we
represent as r, then we have
Wiener Filtering:
The Wiener filtering is optimal in conditions of the
mean square error (MSE). In other words, it
minimizes the generally mean square error in the
development of inverse to remove and noise
smoothing. The Wiener filtering is a linear inference
of the new image. The advance is based on a
stochastic frame. The orthogonality principle implies
that the Wiener filter in Fourier domain. To complete
the Wiener filter in perform we have to
approximation the power spectra of the original
image. For noise is remove the power spectrum is
equal to the variation of the noise. To estimate the
power range of the original image many methods can
be used. A through estimate is the period gram
estimate of the power spectral density (PSD).
C. Energy
For calculating the energy coefficient, the image is
subjected to a wavelet decomposition using the
Daubechies wavelet for up to 5 levels. The wavelet
decomposition involves the image with a low-pass
filter for generating the approximation coefficients
and a high pass filter for generating the detail
coefficients, followed by a down-sampling. The data
image for each level is taken as the approximation
image for the previous level. Another related use is
in image transforms: for example, the DCT transform
(basis of the JPEG compression method) transforms a
blocks of pixels (8x8 image) into a matrix of
transformed coefficients; for distinctive images, it
results that, while the original 8x8 image has its
energy regularly distributed among the 64 pixels, the
182
INTERNATIONAL CONFERENCE ON CURRENT INNOVATIONS IN ENGINEERING AND TECHNOLOGY
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT
ISBN: 378 - 26 - 138420 - 5
www.iaetsd.in
6. changed image has its energy determined in the
lower-upper "pixels”, The decomposition operation
generates the approximation coefficientsA5 and
detailed coefficients B5,B4,B3,B2,B1 as shown in
below.
Ea=∑ (A5)/∑ (B5+ B5+B4+B3+B2+B1)
The final feature vector is taken as the complex
formed of the above three components viz.
F= {Cc, En, Ea}
Classification is done by mapping the feature vectors
of a training set and a testing set into appropriate
feature spaces and calculating differences using
Manhattan distance.
I
Figure 3: Wavelet decomposition of an image
E. Manhattan Distance
The distance between two points in a grid base on a
firmly horizontal and/or vertical path (that is, along
the grid lines), as distinct to the diagonal or "as the
crow flies" distance. The Manhattan detachment is
the plain sum of the horizontal and vertical works;
whereas the diagonal span might be computed by
apply the Pythagorean Theorem. The formula for this
distance between a point X=(X1, X2, etc.) and a point
Y= (Y1, Y2, etc.) is:
Where n is the number of variables, and Xi
And Y is the values of the ith variable, at points X
and Y correspondingly. i The Manhattan Distance is
the distance between two points considered along
axes at right angles. The name alludes to the grid
explain of the streets of Manhattan, which cause the
straight path a car could take between two points in
the city. For the 8-puzzle if xi(s) and y(s) are the x
and y coordinates of tile i in state s, and if upper line
(xi) and upper-line (yii) are the x and y coordinates of
tile i in the goal state, the heuristic is:
ALGORITHM FOR CALCULATING
STATISTICAL TEXTURE FEATURES
Input: Query image for which statistical features has
been computed.
Output: feature vector
1. Calculate Entropy for query image (En) using -sum
(p.*log2 (p)) formula
2. Apply wiener filter for query image and then
calculate correlation coefficient (CC) for query image
and filtered image
3. Apply 5 level decomposition for input query image
and calculate energy for coefficients (Ea)
4. Calculate feature vector F for query image by
using En, Ea, and CC.
Then compare feature vector F of query image with
the database image and if features are equal then the
image is matched.
183
INTERNATIONAL CONFERENCE ON CURRENT INNOVATIONS IN ENGINEERING AND TECHNOLOGY
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT
ISBN: 378 - 26 - 138420 - 5
www.iaetsd.in
7. 4. EXPERIMENTAL RESULTS
The proposed algorithm has been participated in
FVC2004. In FVC2004, databases are more difficult
than in FVC2000/FVC2002 ones. In FVC2004, the
organizers have particularly insisted on: distortion,
dry, and wet fingerprints. Especially in fingerprints
database DB1 and DB3 of FVC2004, the distortion
between some fingerprints from the same finger is
large. Our work is to solve the problem of distorted
fingerprint matching, so the evaluation of the
proposed algorithm is mainly focused on DB1 and
DB3 of FVC2004. The proposed algorithm is also
compared with the one described by Luo et al. and
the one proposed by Bazen et al
Fig.4. Experimental results of the proposed algorithm on
102_3.tif and 102_5.tif in FVC2004 DB1. The images have
been cropped and scaled for view. (a) 102_3.tif. (b)
Enhanced image of 102_3. (c)102_5.tif. (d) Enhanced
image of 102_5. The similarity of these two fingerprints is
0.420820.
Fig.5. Experimental results of the proposed algorithm on
103_2.tif and 103_4.tif in FVC2004 DB3. The images have
been scaled for view. (a) 103_2.tif. (b) Enhanced image of
103_2. (c) 103_4.tif. (d) Enhanced image of 103_4. The
similarity of these two fingerprints is 0.484 111.
5. CONCLUSION
This paper has proposed a quick and efficient
technique of fingerprint recognition using a set of
texture statistical based features. The features are
derived from a correlation coefficient, an entropy
coefficient and an energy coefficient. The features
can be calculated by using fingerprint miniature
points. Moreover such texture based by using color
finger print images. The fingerprint images may be
divided in to separation of red, green and blue
components. And output part combine true color
components. Future work would involve combining
color and shape based techniques to study whether
these can be used to improve recognition rates.
184
INTERNATIONAL CONFERENCE ON CURRENT INNOVATIONS IN ENGINEERING AND TECHNOLOGY
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT
ISBN: 378 - 26 - 138420 - 5
www.iaetsd.in
8. 6. REFERENCES
[1]. A. Lanitis, “A Survey of the Effects of Aging on
Biometric Identity Verification”,
DOI:10.1504/IJBM.2010.030415
[2]. A. K. Jain, A. Ross and S. Prabhkar, “An
Introduction to Biometric Recognition”, IEEE
Transactions on Circuits and Systems for Video
Technology, special issue on Image and Video –
Based Biometrics, 14, 2004, pp. 4-20.
[3]. A. K. Jain, A. Ross and S. Pankanti, “Biometrics:
A Tool for Information Security”, IEEE Transactions
on Information Forensics and Security. 1, 2000.
[4]. A. K. Jain and A. Ross, “Fingerprint Matching
Using Minutiae and Texture Features”, Proceeding of
International Conference on Image Processing
(ICIP), 2001, pp. 282-285.
[5]. A. K. Jain, L. Hong, S. Pankanti and R. Bolle,
“An Identity-Authentication System using
Fingerprints”, Proceeding of the IEEE. 85, 1997, pp.
1365-1388.
[6]. D. Maltoni, D. Maio, A. K. Jain and S. Prabhkar,
Handbook of Fingerprint Recognition.
[7]. S. Chikkerur, S. Pankanti, A. Jea and R. Bolle,
“Fingerprint Representation using Localized Texture
Features”, The 18th
International Conference on
Pattern Recognition, 2006.
[8]. A. A. A. Yousiff, M. U. Chowdhury, S. Ray and
H. Y. Nafaa, “Fingerprint Recognition System using
Hybrid Matching Techniques”, 6th IEEE/ACIS
International Conference on Computer and
Information Science, 2007, pp. 234-240.
[9]. O. Zhengu, J. Feng, F. Su, A. Cai, “Fingerprint
Matching with Rotation-Descriptor Texture
Features”, The 8th International Conference on
Pattern Recognition, 2006, pp. 417-420.
[10]. A. K. Jain, S. Prabhkar, L. Hong and S.
Pankanti, “Filterbank-Based Fingerprint Matching”,
IEEE Transactions on Image Processing, 9, 2000,
pp. 846-853.
AUTHORS
K.Phalguna Rao,
completed M.Tech
information technology
from Andhra University
presently Pursuing PhD.
Life member of ISTE. He
is working as Professor in
the Dept of CSE Published
several papers in the
International Journals and International and
national conferences. Attended several
International and national workshops.
Research Interest areas are Data Base
Systems, Network Security, cloud
Computing, Bioinformatics.
M.Charan Kumar
received sree kalahasthi
institute of technology
degree in computer
science engineering from
the Jawaharlal Nehru
technology university
Anantapur, in 2010, and
received the Audisankara
institute of technology M.Tech degree in
computer science engineering from the
Jawaharlal Nehru technology Ananthapur in
2014, respectively. He published one
International journal and participated four
national conferences and participate One
International conference. He worked as
communication faculty for 3 years in Kerala
and Karnataka.
185
INTERNATIONAL CONFERENCE ON CURRENT INNOVATIONS IN ENGINEERING AND TECHNOLOGY
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT
ISBN: 378 - 26 - 138420 - 5
www.iaetsd.in