I.J.

Image, Graphics and Signal Processing, 2017, 7, 36-44

Published Online July 2017 in MECS (http://www.mecs-press.org/)
DOI: 10.5815/ijigsp.2017.07.04

ASIC Implementation of Finger Print

Recognition Using Overlap-Add and Integer
Wavelet Transform Methods
Shashidhara H. R
Department of Electronics and Communication
JSS Academy of Technical Education, Visvesvaraya Technological University
Bengaluru-Karnataka, 560060, India
Email: [email protected]

Aswatha A. R
Department of Telecommunication
Dayananda Sagar College of Engineering, Visvesvaraya Technological University
Bengaluru-Karnataka, 560078, India
Email: [email protected]

Abstract—The field of fraud identification is reaching a refers to the automated method of verifying a match
very high proportion in the society, thus leading to an between human fingerprints.
increase in the need for fingerprint-based identification. The analysis and study of fingerprints for matching
This paper presents ASIC implementation of fingerprint compares the several print pattern features. This
recognition based on Overlap-add method and Integer fingerprint contains patterns like ridges and minutia
Wavelet Transforms. In overlap-add method, the present points, which are unique features of patterns. A
output overlaps the next output and in the integer to fingerprint is represented by location, types, and some
integer wavelet, low component at 2nd level attributes like angle of inclination of minutiae etc., are
decomposition is taken as approximate integer value. The shown in Fig 1. One hundred years of study on
implementation presents an analysis for speed, area and fingerprints ensures the uniqueness of minutiae based
power dissipation between the two algorithms and other representation for a very large population in the world.
methods. To establish the identity, 50 to 150 minutiae’s are enough
but in automated systems 10 matching minutiae are
Index Terms—Biometrics, Fingerprint Identification, assumed to be sufficient.
Overlap-add Method, Integer Wavelet Transform,
Application Specific Integrated Circuits.

I. INTRODUCTION
Biometrics is the science and technology of analyzing
and measure biological data. Biometric methods of
identification are referred to traditional methods mainly
for two reasons:

 At the time of verification, the physical presence of

the identifying person is required.
 Identification based on biometrics avoids the need
Fig.1. Fingerprint Features
to remember the password or carrying a token.

Fingerprints, palm-print, Iris etc., are some types of

biometrics based method to identify individual person to
II. RELATED WORK
verify their identity. Fingerprint recognition is the most
reliable method for person identification and plays an One of the previous methods for fingerprint
important role in criminal investigation, terrorist identification is Henry Classification System [1], which
identification and national security issues. The impression creates the primary groupings based on fingerprint pattern
of the fingerprint contains friction ridges. They are, types from the logical categorization of ten-print
unique in nature, detailed, difficult to alter and durable fingerprint records. Over the years, Principal Component
for the life span of human being. Fingerprint recognition Analysis (PCA) was an efficient approach in the pattern

Copyright © 2017 MECS I.J. Image, Graphics and Signal Processing, 2017, 7, 36-44
 ASIC Implementation of Finger Print Recognition Using Overlap-Add and Integer Wavelet Transform Methods 37

recognition [2]. This approach effectively extracts the scale value. Hence in thresholding is performed based on
global features from the aligned fingerprint images. PCA threshold values. First, the image is divided into number
transforms a number of correlated variables into of blocks and for each block grey-scale mean is
uncorrelated variables known as principal components. calculated. If the grey value (Gi) of a particular pixel is
Anil K. Jain [3] in his paper has proposed filter bank less than the mean of that block (Bi), then zero is
based algorithm for fingerprint recognition. Gabor filters assigned otherwise one is assigned. The outcome of this
are used to consider the global and local details of process is shown in Fig 3, is a binary image having two
fingerprint. Anil K. Jain, et. al,[4] have proposed a levels of information.
method on the combination of different fingerprint
matching algorithms to develop the performance of a I(x; y) = 0; if Gi < Bi; = 1; otherwise; (1)
Finger Print Recognition (FPR) system. In this paper they
B. Skeletonisation
have combined the scores values from logistic transform,
which are generated from three dissimilar fingerprint Skeletonisation of binary images is done by a process
matching algorithms. Arun Ross, Anil K Jain and Salil called thinning that reduces all lines to a single pixel
Prabhakar [5] have proposed fingerprint matching using thickness. Thinning algorithm retains the connectivity of
minutiae and texture features. F. G. Hashad, et. al. [7] ridge structures while forming a skeletonised finger
have introduced Mel Frequency Cepstral Coefficients image. To eliminate a pixel the following conditions must
(MFCCs) for Fingerprint Recognition. Manisha be considered:
Redhuand and Dr.Balkishnan [9] has proposed new
method in FPR system. They used score values obtained  The pixel will not be reasoned as an endpoint.
from matching method and image enhancement is done  The elimination of pixel does not break
by using histogram technique and FFT. S. Gayathri and connections of the skeleton.
Dr. V.Sridhar [12] have shown the ASIC implementation  The elimination does not cause excessive erosion
of fingerprint recognition process. The efficiency of the of the region.
fingerprint recognition system depends on the minutiae
obtained from the unprocessed image.
The objective of this paper is to design a high speed,
low area based fingerprint recognition system, which can
process every fingerprint image of the user. The finger
print identification has been implemented by using two
algorithms: they are Fast Convolution and Integer
Wavelet Transforms. The fast-convolution method uses
overlap add method than overlap save method, because
overlap save method saves previous stage outputs, thus
requiring huge memory. In the overlap add method the
previous stage is not saved and hence memory
requirement will be less. The integer wavelet is a 2nd
level transformation technique to transform the low order
coefficient in the time-frequency domain. With an On-
chip architecture being developed for a high speed and
low area FPR system, the designed system is verified as
Fig.2. Flow Diagram for Fingerprint Recognition
an IP for Functionality and Performance. The rest of the
paper is organized with Section II describing the overlap-
add method and IWT algorithm used. Section III presents
simulation results and performance evaluation of the
systems. Section V gives the conclusion for the paper.

III. METHODOLOGY
The flow diagram shown in the Fig 2 explains general
process of fingerprint recognition. The
thresholding/binarization and skeletonisation are the pre-
processing steps and feature extractions are common to
both the algorithms. Features of fingers are extracted
Fig.3. Binarized Image
using fast convolution and IWT method after which, bit
wise matching is done using hamming distance. In the Fig 4 is the skeletonized image used in the
A. Thresholding/Binarisation subsequent extraction of minutiae. After Skeletonisation
the features like ridges, delta points are extracted using
In fingerprint image the foreground have high grey Overlap-add and IWT method, as explained in the
value and the background regions exhibit a low grey- following sections.

Copyright © 2017 MECS I.J. Image, Graphics and Signal Processing, 2017, 7, 36-44
38 ASIC Implementation of Finger Print Recognition Using Overlap-Add and Integer Wavelet Transform Methods

First the input signal f is split into smaller disjoint tiles

C. Feature Extraction
f1; f2; ::; fm. Therefore f(n) = f1(n) + f2(n) + :. For the
computation of convolution the selected signal fi and the
i. Overlap add method large kernel g is considered. That is,
The overlap-add method is a type of convolution for
fixed coefficients and it is computationally efficient but h(n) = f(n)*g(n); f(n) = f1(n)+f2(n)+f3(n)+::::f (n)*g(n) =
requires more memory. This method can be applied to (f1(n)+f2(n)+f3(n)+::::)*g(n)h(n) (2)
frequency domain and time domain using DFT and IDFT
techniques. In basic convolution technique, after the Let us consider a input block of size L points and the
multiplication the output values are added. The final size of DFT and IDFT will be N = L+M-1. To compute N
output will be very long sequence, to computation of point DFT concatenate each block with M-1 zeros.
these long sequence is very difficult. In overlap add Therefore the data block will be,
method the output from the rear end of the former block
is overlapped with the front end of the present block. The x1(n) = x(0); x(1); ; x(L-1); 0; 0; :; 0x2(n) = x(L);
process of overlapping and addition of the convoluted x(L+ 1); ; x(2L-1); 0; 0; :; 0x3(n) (3)
signal is carried out. Hence the name overlap-add method.
In this work overlap add method is used for feature & So on.
extraction. The Fig 5 shows overlap-add method Therefore 2N point DFT is,
procedure, the input signal is f and it is split into small
signal. Then it is convoluted with the kernel g. Each Ym(k) = H(k)Xm(k) k = 0; 1; ::;N-1 (4)
convoluted blocks are overlapped and then added. Finally
we get the convoluted sequence. The length of this The IDFT having data block of length N are free from
sequence obtained is less than the basic convolution aliasing, the sequence of DFT & IDFT is increases to N
method. points by concatenating zeros. Each block is terminated
by M-1 zeros, the last M-1 points from each output block
is overlapped and added to the 1st M-1 points of the
immediate block.

That is,

Y (n) = y1(0); y1(1); :y1(L-1); y1(L) + y2(0);

y1(L + 1) + y2(1); ::; y1(N-1) + y2(M-1); y2(M); (5)

Overlap add is a linear convolution method between

the infinitely long input signal x(k) & the finite impulse
response Ki as shown in Fig 5. The following points are
explained by the Fig 5. Fig 5 overlap add method in
Fig.4. Skeletonized Image frequency domain

 Add the impulse response Ki with N zeros in

window g & transform to frequency domain with
Kl(m) with l=0,1,. . . , 2N-1, follows for every
iteration m=0,1,2,. . .
 Choose N samples from signal x(m) and add these
by N zeros in segment to produce a block
 Transform to frequency domain which results in
the frequency bins Xl(m) with l=0, 1, .2N-1.
 Convolute Xl(m) & Wl(m) for l=0, ....2N-1.
 Transform back convoluted values to time domain
using IDFT ei(m) with i=0, 1,....2N-1.
 Perform desegmentation to generate the output
signal with k=mN+I calculated as The algorithm
of overlap add method is,
 Break the input signal f (n) into non-overlapping
blocks fm(n) of length L.
 Pad zero for g(n) to be of length N = L + M - 1.
 Take N-DFT of g(n) to give G(k), k = 0,1,. . .N-1.
 For each block m:
Fig.5. Overlap add method

Copyright © 2017 MECS I.J. Image, Graphics and Signal Processing, 2017, 7, 36-44
 ASIC Implementation of Finger Print Recognition Using Overlap-Add and Integer Wavelet Transform Methods 39

o Zero pad fm(n) to be of length N = L + M convolution based DWT. In IWT the computational
-1. complexity is reduced to almost a half as compared to
o Take N-DFT of fm(n) to give Fm(k), k = convolution approach.
0,1,....N -1. The main advantages are as follows:
o Multiply: Hm(k) = Fm(k) * G(k), k =
0,1,....N -1.  It consents to a faster implementation.
o Take N-IDFT of Hm(k) to give Hm(n), n =  It permits fully in place calculation, in other words,
0,1,....N -1. no auxiliary memory is needed and the original
signal (image) can be replaced with its wavelet
 Form h(n) by overlapping the last M -1 samples of transform.
hm(n) with the rest M -1 samples of hm+1(n).  By reversing the operations, the inverse wavelet
 The adding overlapped samples of hm(n) and transform is created.
hm+1(n).
 The 5 DFT overlap add method is shown in Fig 6. There are several frameworks are explained, in that
most popular are S+P, Lifting scheme and overlapping
Fig 6 Overlap Add Method with 5 DFTs in frequency rounding transform (ORT) frameworks. The S-transform
domain. is the most basic framework for ITI transform. The below
section briefly describes the S-transform for the basics of
ii. Integer wavelet transform
ITI and also other three frameworks which are generally
The second algorithm is used to extract the finger used.
image feature is Integer Wavelet Transform (IWT). The S-Transform: The Sequential (S) transform is
disadvantage of DWT is that it contains real numbers as uncomplicated and ubiquitous ITI transform. The S
wavelet coefficients. So efficient lossless coding is not transform is the approximated (linear) Haar transform of
possible using linear transforms. The lifting scheme (LS) ITI and is allied with UMD filter bank as shown in Fig 7.
is an efficient implementation of the DWT and also The forward and inverse transform is calculated by the
perfect reconstruction is ensured. This creates new analysis and synthesis side of the filter bank respectively
transformation, like integer wavelet transform (IWT). as shown in Fig 7. The mathematical analysis of the S-
IWT is a basic modification of linear transforms, in transform is given by,
which output is rounded to the nearest integer. It is also
of interest for hardware implementations, where the use Y0 [n]=[1/2(x[2n]+x[2n+1])] (6)
of floating point is still a costly operation.
and

Y1 [n]=[1/2(x[2n]-x[2n+1])] (7)

Where the forward transform divides the input signal

x[n] into low y0[n] and high y1[n] pass components. The
above forward transform are considered for infinite
length and for finite length the symmetric and periodic
extension can be considered. The S-transform shows two
key remarks, first any two coefficient values are
calculated from their sum and difference. Second, the
sum and difference of any two coefficients have same
parity (which omits the fractional part without losing
information).
S+P transform: This transform is known as sequential
plus prediction, it is a direct extension of S-transform.
The framework of S+P transform is shown in Fig 7,
mathematically it can be expressed as

Y0 [n]=[1/2(x[2n]+x[2n+1])] (8)

and

Y1 [n]=[v_0[n]-[t[n]+1/2]] (9)
Fig.6. Overlap add method
y0 [n]=[1/2(x[2n]-x[2n+1])] (10)
The convolution based DWT needs a large number of
arithmetic computations and a large memory for storage. and
The IWT requires fewer computations compared to the

Copyright © 2017 MECS I.J. Image, Graphics and Signal Processing, 2017, 7, 36-44
40 ASIC Implementation of Finger Print Recognition Using Overlap-Add and Integer Wavelet Transform Methods

L0, L1 and K are integers satisfying L0 ≤ L1 and K≤ 1. forward transform of 1-D two channel UMD filter bank.
The forward transform divides the input coefficients The cascade of two input two outputs ITI networks is
x[n] into low-pass and high-pass coefficients as y0[n] and performed from polyphase filtering.
y1[n] respectively.
From the above equations, it is found that the S+P
transform is generated by adding an extra step to S-
transform; that is the high-pass component of S-transform
is adjusted with an extra prediction operation. The
predictor component αi and βi is shown in table 1.

Table 1. Predictor Co-efficient

Predictor α-1 α0 α1 β-1 Fig.9. Overlapping Rounding Transform

A 1/4 1/4 0 0
Finger features are extracted by using the lifting
B 3/8 2/8 0 0
scheme based ITI transform.
C 8/16 4/16 1/16 6/16
D. Minutiae Reduction
Minutiae are the trivial details of fingerprint. An
unavoidable source of error in fingerprint recognition is
false minutiae. These are inevitable because of distortion
such as scars, sweat, etc. Existence of false minutiae
forces the matching process to fail. It is possible to handle
excessive numbers of them; therefore after the feature
extraction phase, the minutiae list is analyzed to eliminate
false minutiae. Our approach is to eliminate minutiae
features using distance criteria, e.g., minutiae which are
too close to each other are discarded. For a particular
Fig.7. S+P transform with UMD filter bank minutiae point within the D8 distance (also called as
chessboard distance), only one minutia point is
Lifting framework: The most popular framework of ITI considered. Any other minutiae point existing within of 3
transform is based on the lifting scheme. The lifting are all eliminated.
framework is a polyphase comprehension of UMD filter
banks, which adopts ladder networks for filtering. The 1- E. Matching using Hamming Distance
D two channel UMD filter bank contains 2 steps, they are The Hamming distance between two strings of equal
lifting and scaling steps. In lifting step, the filters Ak is length is calculated by creating a sequence of logical
constituted and in scaling step the amplifiers of gain sk as Boolean values indicating matches and mismatches
shown in Fig 8. between their position in the two inputs, and then
The ladder network of lifting scheme has to sustain summing the sequence.
their invariability even for quantization error (rounding
error by finite precision arithmetic). The construction of I(x,y)=I(xi,yi) xor I(xj,yj) (11)
lifting scheme based ITI transform follows initially to
constrict the lifting scheme, eliminate the scaling function
having a non-integer gain factor, and then transforms
IV. RESULT
each lifting step to the rounding operation Q at the output
of its corresponding filter as shown in Fig 8. The inputs which are described in the modules are
functionally verified using ModelSim. After the
successful completion of simulation, the design is
synthesized using Cadence tool.
The input image is in matrix form. Image enhancement is
done for the images to improve the quality \& mean is
calculated. If the grey value of a particular pixel is less
than the mean of that block then zero is assigned,
otherwise one is allotted. The outcome is an binary image
containing two logical levels of information.
Skeletonisation is done using thinning process to reduce
all lines to single pixel values. The skeletonized image is
used for feature extraction, which is done using overlap
Fig.8. Lifting framework with UMD filter bank
add method of fast convolution and IWT. Then the
Overlapping Rounding Transform: Overlapping minutiae reduction is done to remove false minutiae.
Rounding Transform (ORT) framework is a polyphase Finally matching of the test fingerprint with the reference
apprehension of UMD filter bank. The Fig 9 shows the images is done.

Copyright © 2017 MECS I.J. Image, Graphics and Signal Processing, 2017, 7, 36-44
 ASIC Implementation of Finger Print Recognition Using Overlap-Add and Integer Wavelet Transform Methods 41

In Fig 12 the matched condition is shown, when both

A. Fast Convolution Method
the test and reference finger images are same, the match
Fig 10 shows the top module of Overlap-add Method, signal is 1.
if the test image is matched with the reference image. It
shows the end point branch points and matching score.

Fig.10. Finger Print Top Module for overlap add method

Fig.12. Simulation Results for Matched Condition

Table 2 describes the signal details of the top module. The resultant cell layout is shown in Fig 13. The Table
3 and Table 4 describe the memory usage and power
Table 2. Signal Description of Top Level Module.
utilization before routing. The total area of 9100.560µm2,
Signal I/O Description the total power 444030.394nW and total cells of 1496 in
CLK Input Synchronous clock signal of finger print top that sequential gate of 1211 and the logic gate count of
module 743.
rst Input Reset signal
Table 3. Area Utilization Summary of overlap add Method
en Input Enable signal
MATCH Output The final authentication match signal Instance No. of Cells Cell Area
Sequential 1211 8036.582µm2
Table 2 describes the signal details of the top module. Inverter 212 212.420 µm2
Logic 743 851.558 µm2
The simulation results for the finger print top module are Total Area 1496 9100.560 µm2
given in Fig 11 and Fig 12. In this module the two finger
images are obtained from the text file. All these data are Table 4. Power Utilization Summary of overlap add Method
of eight bit in nature. The module operates for the
Leakage Power 291.522 nW
positive edge of clk and rst=0. Initially when the clk Dynamic Power 443738.872 nW
signal is at the positive edge and rst=1, no data Total power 444030.394 nW
evaluation process is taking place as shown in Fig 11 and
Fig 12. Fig 11 shows the unmatched condition if the test
image does not match with the reference image. It shows
that fingerprint is not identified.

Fig.11. Simulation Results for Unmatched Condition Fig 13. Final routed Cell layout of the overlap add Method

Copyright © 2017 MECS I.J. Image, Graphics and Signal Processing, 2017, 7, 36-44
42 ASIC Implementation of Finger Print Recognition Using Overlap-Add and Integer Wavelet Transform Methods

B. Integer Wavelet Transform

The top module of Integer Wavelet Transform as
shown in Fig 14. Table 5 describes the signal details of
this top module.

Fig.15. Simulation Results for No-Match Condition

Table 8 shows the comparative study of IWT, Overlap-

add, O2D and Image enhancement technique compared
Fig.14. Finger Print Top Module for Integer Wavelet method
between area, gate count, power dissipation and delay.
The IWT technique shows better performance with
respect to area, power and gate count. Fig 18 shows the
Table 5. Signal Description of IWT Top Level Module comparative study of recognition rate between IWT and
Overlap-add technique, IWT shows higher recognition
Signal I/O Description
rate than Overlap-add method.
CLK Input Synchronous clock signal of finger print top
module
rst Input Reset signal
en Input Enable signal
Match_t Output The final authentication match signal

The simulation results for the finger print top module

are shown in Fig 15 and Fig 16. The module operates for
positive edge of clk and rst signal i.e, when rst is 1 the
data is considered else the module resets to zero. Fig 15
shows the unmatched condition if the test image does not
match with the reference image. It shows that fingerprint
is not identified with match as zero.
In Fig 16 the matched condition is shown, when both
the test and reference finger images are same, the match
signal goes to 1.
The cell layout as shown in Fig 17 and Table 6 and
Table 7 shows the area and power dissipation of IWT Fig.16. Simulation Results for Match Condition
based finger print module. The total area of 53271 µm2
having 8752, sequential cells of 912 has 23701.104 µm2
of area, the inverter having 158 cells and 334 µm2 area.
The total power dissipation of 388021.953nW in that
leakage power of 285.735nW and dynamic power
387736.218nW. Table 8 shows the area, number of cells
and gates after detailed routing.

Copyright © 2017 MECS I.J. Image, Graphics and Signal Processing, 2017, 7, 36-44
 ASIC Implementation of Finger Print Recognition Using Overlap-Add and Integer Wavelet Transform Methods 43

Fig.18. GAR vs FAR of overlap add and IWT

Fig.17. Final routed Cell layout of the Integer wavelet Design

V. CONCUSION
Table 6. Area Utilization Summary of Integer Wavelet Method The comparative study reveals that IWT is highly
Instance No. of Cells Cell Area efficient compared to overlap-add Method, Image
Sequential 912 23701.104µm2 enhancement technique and O2D. The characteristics of
Inverter 158 334.454 µm2
low area, low power and low delay for IWT is shown
through the ASIC Implementation of the same. It has
Logic 682 20235.125 µm2
been observed that the recognition rate of IWT is better
Total Area 1752 44270.683 µm2
than overlap-add method. The credibility to use both
algorithms has been justified with improved recognition
Table 7. Power Utilization Summary of Integer Wavelet Method rate, area, power and delay, compared to other algorithms.
Leakage Power 285.735 nW In future, only those features in the required region of the
Dynamic Power 387736.218 nW finger can be concentrated upon and can be derived for
higher recognition rate and effective ASIC
Total power 388021.953 nW
implementation.

Table 8. Comparison of device parameters between different algorithms

Overlap- Add IWT O2D[16] Image Enhancement [12]
Technology/FPGA Tsmc18.1.0 Tsmc18.1.0 Vertex-xc5vlx110t Tsmc18.1.0
Area/LUT/Register-FF 8888.141 µm2 6852.2 µm2 94521 70242.49 µm2
Sequential 1211 912 --- 1744
Gates Unresolved 212 158 --- 414
Logic 743 682 --- 6183
Delay/Speed 1565ps 1032ps 56.686MHz (Clock rate) ---

[5] Anil Jain, Arun Ross and Salil Prabhakar, “Fingerprint

REFERENCES matching using minutiae and texture features”, Appeared
in Proc. of Int’l Conference on Image Processing (ICIP),
[1] Cheong Hee and Haesun Park, “Fingerprint classification Thessaloniki, Greece, Oct 7 - 10, 2001, pp.282-285.
using fast Fourier transform and nonlinear discriminant [6] Wang Yongxu, Ao Xinyu, Du Yuanfeng and Li Yongping,
analysis”, Journal of Pattern Recognition, Elsiever, 2005, “A Fingerprint Recognition Algorithm Based on Principal
Volume 38, Issue 4, pp.495-503. Component Analysis”, 2006 IEEE Region 10 Conference
[2] M. Kawagoe and A. Tojo, “Fingerprint Pattern (TENCON 2006), pp.1-4.
Classification”, Pattern Recognition, Volume 17, Issue 3, [7] Hashad, F. G., “A new approach for fingerprint
1984, pp.295-303 recognition based on Mel frequency cepstral coefficients”,
[3] Anil K. Jain, Salil Prabhakar, Lin Hong, and Sharath Proc. IEEE International Conference on Computer
Pankanti, “Filter bank-Based Fingerprint Matching”, Engineering \& Systems, 2009, pp.263-268.
IEEE Transactions on Image processing, Volume 9, Issue [8] Ramandeep Kaur, Parvinder S. Sandhu, and Amit Kamra,
5, 2000, pp.846-859 “A novel method for fingerprint feature extraction”, Proc.
[4] Anil K.Jain, Salil Prabhakar and Shaoyun Chen, IEEE International Conference on. Networking and
“Combining multiple matchers for a high security Information Technology, 2010, pp.1-5.
fingerprint verification system”, Pattern Recognition [9] Manisha Redhu and Dr.Balkishan, “Fingerprint
Letters, Elsivier, 1999, Volume 20,Issue 11, pp.1371- Recognition Using Minutiae Extractor”, International
1379.

Copyright © 2017 MECS I.J. Image, Graphics and Signal Processing, 2017, 7, 36-44
44 ASIC Implementation of Finger Print Recognition Using Overlap-Add and Integer Wavelet Transform Methods

Journal of Engineering Research and Applications (IJERA) Signal Processing(IJIGSP), Vol.9, No.2, pp.9-18,
ISSN: 2248-9622, Volume 3, Issue 4, 2013, pp.2488-2497. 2017.DOI: 10.5815/ijigsp.2017.02.02
[10] Ajit Kumar, and Cyril Kwong, “Towards contactless, low- [20] Iwasokun Gabriel Babatunde, “Fingerprint Matching
cost and accurate 3D fingerprint identification”, IEEE Using Minutiae-Singular Points Network”, International
Transactions on Pattern Analysis and Machine Journal of Signal Processing, Image Processing and
Intelligence, Volume 37, Issue 3, 2015, pp.681-696. Pattern Recognition”, Volume 8, Issue 2, 2015, pp.375-
[11] M. Kamaraju and P. A. Kumar, “DSP based embedded 388.
fingerprint recognition system”, Proc. IEEE 13th
International Conference on In Hybrid Intelligent Systems
(HIS), 2013, pp.6-11.
[12] S. Gayathri and Dr. V.Sridhar, “ASIC Implementation of Authors’ Profiles
Image Enhancement Technique for Fingerprint
Recognition Process:, Proc. International Conference on Shashidhara H R is working as Assistant
Contemporary Computing and Informatics, 2014, pp.868- Professor in the Department of Electronics
873. and Communication Engineering, JSS
[13] S. Santhosh, “Design and development of a security Academy of Technical Education,
module with inbuilt neural network methodologies and an Begaluru and research scholar in the
advanced technique on fingerprint recognition”, IEEE Department of Electronics and
International Conference on Circuit, Power and Communication Engineering at Dayanand
Computing Technologies, 2014, pp. 1490-1495. Sagar College of engineering affiliated to
[14] Fred Kaggwa, John Ngubiri and Florence Tushabe, Visvesvaraya Technological University. He obtained his M.
“Evaluation of multiple enrollment for fingerprint Tech degree in VLSI Design and Embedded System from
recognition”, IEEE Global Summit on Computer \& Visvesvaraya Technological University in the year 2008 and
Information Technology, 2014, pp.1-6. Bachelor of Engineering in Electronics and Communication
[15] BinduGarg, ArjunChaudhary, KunalMendiratta and Vijay from Mangalore University in the year 2001. His major area of
Kumar, “Fingerprint Recognition Using Gabor Filter”, research On-Chip and ASIC implementation of Image
International Conference on Computing for Sustainable processing techniques, TLM architecture and Verification
Global Development (INDIACom), 2014, DOI: Methodologies, Network on chip design and implementation,
10.1109/IndiaCom.2014.6828105. Biometric and Image processing. He is Life member of IETE.
[16] Satish S Bhairannawar, SayantamSarkar, Raja K B and
Venugopal K R, “An Efficient VLSI Architecture for
Fingerprint Recognition using O2D-DWT Architecture Dr. Aswatha A R is working as Dean-
and Modified CORDIC-FFT”, IEEE International Academics, Professor and Head of the
Conference on Signal Processing, Informatics, department of Telecommunication at
Communication and Energy Systems (SPICES 2015), Dayanand Sagar College of Engineering.
Kozhikode (Calicut), India, 2015, pp.193-197. Lastly (in 2010), he received Ph.D. in the
[17] Om Preeti Chaurasia, Saumya Ranjan Giri, Anchal Garg, area of VLSI Circuit Design, and also he
"A Novel Algorithm for Minutiae Matching", earned M. Tech in Lighting Science and
International Journal of Image, Graphics and Signal Engineering in 1996 and MS in Software
Processing(IJIGSP), Vol.4, No.3, pp.8-14, 2012.DOI: Engineering in 2002 and also Engineer degree in Electricals and
10.5815/ijigsp.2012.03.02. Electronics in 1991. He is contributed IGI-Global book chapter
[18] Jyoti Malik,Dhiraj Girdhar,Ratna Dahiya,G. and published over 27 papers in referred international journal
Sainarayanan,"Reference Threshold Calculation for and 30 International conferences. His major area of research is
Biometric Authentication", IJIGSP, vol.6, no.2, pp.46-53, ASIC and On-Chip architecture implementation and verification
2014.DOI: 10.5815/ijigsp.2014.02.06 methodologies, Low power circuits, biometrics, pattern
[19] Shiv Gehlot, John Deva Kumar,"The Image Segmentation recognition and image processing. He is a member of the IEEE,
Techniques", International Journal of Image, Graphics and Member of IE (INDIA) and Life member of ISTE.

How to cite this paper: Shashidhara H. R, Aswatha A. R,"ASIC Implementation of Finger Print Recognition Using
Overlap-Add and Integer Wavelet Transform Methods", International Journal of Image, Graphics and Signal
Processing(IJIGSP), Vol.9, No.7, pp.36-44, 2017.DOI: 10.5815/ijigsp.2017.07.04

Copyright © 2017 MECS I.J. Image, Graphics and Signal Processing, 2017, 7, 36-44