Special Issue - 2015 International Journal of Engineering Research & Technology (IJERT)

ISSN: 2278-0181
NCETET-2015 Conference Proceedings

Indian Currency Detection for the Blind using

MATLAB

Alen Thomas Varghese Jewel Domonic Savio Antony

Dept.of Electronics & Communication Dept.of Electronics & Communication
AmalJyothi College of Engineering AmalJyothi College of Engineering
Kanjirapally, India Kanjirapally, India

Jacob Shibu Chacko Jerrin Varghese

Dept.of Electronics & Communication Dept.of Electronics &Communication
AmalJyothi College of Engineering AmalJyothi College of Engineering
Kanjirapally, India Kanjirapally,India

Abstract—This paper has led to the development of the the recognition of the Indian currency both banknotes and
prototype of a system that recognizes the denomination of all coins, by affixing camera with their goggles.
India currency i.e. notes and coins, aimed for visually
impaired people. It is capable of reproducing audio messages In order to recognize the notes, in the first stage, a
that announce the denomination of a banknote placed in front
digital processing of the image obtained by the camera has
of a camera by processing each frame of its continuous
filming. This message is heard by the visually impaired been made in order to locate a particular region of interest
through an earphone. This work takes its theoretical basis within that image. Then, taking that region of interest as an
from the Digital Image Processing (DIP) techniques. The input image, a recognition stage is implemented making
recognition of the bank notes is primarily from the image use of the ‘Face Recognition using Eigenfaces method,
recognition method known as Eigen faces, which is based on which, in turn, is based on the mathematical technique
the Principal Component Analysis (PCA) mathematical known as Principal Component Analysis (PCA). Finally,
theory. And in order to recognize the coins of all results of tests made on the system implemented, are
denominations a system has been created which recognizes presented.
coin based on image subtraction technique. The process
performs 3 checks (radius, coarse and fine) on the input image.
The stated subsequent checks enable the technique to endorse The coin recognition is dealt by the method using
Rotation Invariance, thus obviating the need of placing the image subtraction technique. The image subtraction
coin at a certain angle. Subtraction between the input object technique takes two images as input and gives a third
image and database image is performed. Further, plotting the image as output, whose pixel values are simply the pixel
resultant values gives minima which if less than a standard values of the first image minus the corresponding pixel
threshold establishes the recognition of the coin. To establish a values of the second image.. It also incorporates the
distinguished output MATLAB stimulation can be taken into checking of radius which would assist in choosing the
accord. Therefore, from this anticipated method a well- befitting coin from the database. Database amasses the
developed system can brought out to help the visually
standard coin used for recognizing the input image. Once
impaired to overcome their disability so that they can have a
well off life in the society. the precise image is selected, its feature are extracted and
IndexTerms— Indian Currency, Image Processing, PCA, subtracted from the input coin image. Image rotation
Rotation Invariance, Image Subtraction. invariance is introduced by rotating the image at fixed
angular interval thus providing us with the exact angle of
I. INTRODUCTION difference between the coins on analyzing the plot of the
The recognition of currency denomination is among the subtracted values.
most challenging problems faced by visually impaired We then combine these results accordingly and the
people since it constitutes an essential part of their day to required output is reached out towards the visually
day affairs. Though there are many recognition methods for impaired by an audio voice which reaches out to them
them but these methods put them at high risk of becoming through an ear phone.
victims of deception, either because of the complexity and
discomfort of using those methods or due to the dishonesty
of certain people who could try to take advantage of their
visual disabilities during the process. The present paper
describes the phases of the implementation of a system for

Volume 3, Issue 05 Published by, www.ijert.org 1

Special Issue - 2015 International Journal of Engineering Research & Technology (IJERT)
ISSN: 2278-0181
NCETET-2015 Conference Proceedings

II. RECOGNITION OF BANK NOTES W×W window, centered on the pixel (x,y) , where W is 1/8
The Indian Bank Notes are of seven denominations and of the grayscale image’s width and W is a mean
each of it has the value of those denominations can be ponderation factor.
noticed at the corners at the front as well as the back of the
notes. t(x,y)=τη(x,y)(3)
The mean ponderation factor was heuristically chosen
A. LOCATION OF THE REGION OF INTEREST
as η=1.1 because, as a result of this selection, the white
One of the four regions of interest is determined from border of the banknote could be distinguished from the
the image that is taken by the camera, to be then recognized background in the binary image shown in Fig.I.A.2
through the finding of the vertex coordinates of the
banknote'sregion of interest, by using the end to end
projection of the binary, eroded image of the input image.
When such a region is detected, it is cropped to and scaled
to a 60 × 90 pixel resolution, region of interest.
This obtained image is ready to be recognized with the
processes described in section C, and will be known in
there as input image. The various processes involved after
the location of the region of interest are described in this
section.
FIG I.A.2 Binary Image
1) Conversion of RGB Image into Grayscale
The RBG-colour image delivered by the digital camera 3) Erosion Of Binary Image
is transformed into a grayscale image seenof the Fig. I.A.1 As one can see in Fig.A.2, the binary image contains
through the calculation of the average of the three channels noise, which is presented as white pixels isolated in the
RGB of the image, using eqn 1 background. Erosion aims at filtering it with a view to
eliminate that noise to the maximum extent feasible. The
GRAY= (GREEN+RED+BLUE)/3 (1) operation constructs a new binary eroded image be by
applying a structuring element, D to the original binary
image, B, using (4)

𝑏𝑒 (x,y)=𝐵∅𝐷 = {𝑏⃗ ∈ 𝐸 2 |𝑏⃗ + 𝑑 ∈ 𝐵, 𝑑 ∈ 𝐷}

(4)
1 1 1
The structuring element D, which is: D = 1 *1 1
1 1 1

FIG I.A.1 Grayscale Image 4) Calculating the Projection Profile

2) Binarization of the Grayscale Image The vertical PV and horizontal PH projection profile of
the eroded image be (x,y) of Q rows and R columns is
The method implemented was the adaptive thresholding defined as the sum of all white pixels for rows and
using local means. This method proved to be robust to columns using (5) and (6), respectively
strong illumination changes which will allow the system to 𝑃𝑉 =∑𝑅𝑥=1 𝑏𝑒 (x,y);∀𝑦 =
𝑄
work in varied illumination conditions. The basic 1,2, … . 𝑄(5) 𝑃𝐻 =∑𝑦=1 𝑏𝑒 (x,y);∀𝑥 =
methodology of such method consists of calculating for a
grayscale image, g(x,y) , a threshold t(x,y) for each pixel 1,2, … 𝑅(6)
position (x,y) such that the binary image b(x,y) is as in eqn
(2). The outcome of applying the equations (5) and (6) to
the eroded binary image is illustrated in Fig. 2.5. The
0, 𝑔(𝑥, 𝑦) ≤ 𝑡(𝑥, 𝑦) coordinates of the region of interest are those rows and that
b(x,y)={ (2) column whose vertical and horizontal profile respectively
1, 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒 exceeds a minimum threshold.

The threshold, t(x,y) is computed through the equation

(3) using the mean η(x,y) of the pixel intensities in a

Volume 3, Issue 05 Published by, www.ijert.org 2

Special Issue - 2015 International Journal of Engineering Research & Technology (IJERT)
ISSN: 2278-0181
NCETET-2015 Conference Proceedings

5) Cropping & Scaling of the Region of Interst the denomination of the banknote which, at the time of the
recognition, is being filmed by the the smartphone camera.
Once the coordinates of the vertices of the region of
interest are located, such portion is cropped from the
grayscale image to an aspect ratio of 2:3, and subsequently
scaled to a 60×90 pixel resolution using the nearest 1) The Prospectus Of Principle Component Analysis
neighbor interpolation method, as illustrated in Fig.I. A.3
The goal of principle component analysis (PCA) is to
reduce the dimensionality of the data while retaining as
much information (but no redundancy) as possible in the
original dataset.
PCA allows us to compute a linear transformation that
maps data from a high dimensional space to a lower
dimensional sub-space.

2) Setting The File System

As training set samples, M=168 images of banknotes of

different denominations were taken. The samples exhibit
FIG I.A.3 Cropped and Settled Image
the four portions of interest of each denomination (see Fig.
1), and were processed in the same manner as in chapter 1.
6) Normalization of the Scaled Image shows 24 sample images, of resolution 60 × 90. Each
sample is represented in a 5400 × 1 vector, Γi . Next, we
In order to reduce the effects of the illumination
proceed to calculateΨ, the average image of the training
changes on the scaled image g esc , the latter must be
set, using eqn (8).
normalized by applying eqn (7), operation which carries
the value of its current mean η to zero and the value of its
1
current standard deviation σ to one. The input image gN is 𝛹= ∑𝑀
𝑖=0 𝛤𝑖 (8)
𝑀
then obtained and will be classified in the recognizing Each vector Γi is subjected to a process of
phase. normalization. Each corresponding normalized vector is
obtained using eqn (9).
gN= (𝑔𝑒𝑠𝑐 – η)/ σ (7)
𝛷𝑖 =𝛤𝑖 - 𝛹, ∀ 𝑖 =1…M (9)
B) RECOGNITION OF IMAGE The set of normalized vectors is then organized as
columns in a matrix written as A=[𝛷1 𝛷2 … 𝛷𝑀 ] where
It applies the steps for image recognition using every column of A is a normalized vector representing a
eigenfaces which, in general terms, can be called eigen sample image. From A we shall obtain a symmetric matrix
images which is taken in accordance with the principle L, using eqn (10).
component analysis (PCA), since here such approach is
used for the recognition of images other than faces. L= 𝐴𝑇 𝐴 (10)
Following that approach, a vector representation of the According to the theory of PCA, the principal
N-dimensional input image is projected to a new K- components, that form a change of basis transformation
dimensional, reduced space through a change of basis that is suitable to better represent the similarities and
operation. As a result, the input weights vector Ωin , is differences within the original data set, are each
obtained. The new basis is composed of the eigenvectors of eigenvector ui of the covariance matrix of the original data
the original covariance matrix, the one corresponding to the set corresponding to the largest absolute values of the
different banknotes data set. Such eigenvectors are called eigenvalues. These principal components are those known
eigen images. as eigen images and are obtained from each eigenvector of
Finally, the banknote denomination of the input image the L matrix using eqn (11).
is identified by performing a suitable metric among the
input weights vector and each one of the sample weights 𝑢𝑖 =A𝑣𝑖 (11)
vectors that are obtained from the projection of each of the
sample images in the new K-dimensional reduced space, For the foregoing reasons, we chose the first K=24
which from now on, will be called banknotes space. The eigenvectors corresponding to the eigenvalues of the
metric for the identification used in this paper is the largest absolute value from the set obtained using eqn (11).
Mahalanobis distance which makes use of the eigenvalues Since, within PCA, there is no formal method for choosing
of the original covariance matrix. the value of K, we have chosen 24 based on the fact that
Since the system is aimed at visually impaired people, we use 24 image species of different appearance for
the system output is an audio message that communicates recognition (4 regions of interest × 6 denominations).Once

Volume 3, Issue 05 Published by, www.ijert.org 3

Special Issue - 2015 International Journal of Engineering Research & Technology (IJERT)
ISSN: 2278-0181
NCETET-2015 Conference Proceedings

K selected, we normalize (match their module to 1) and used. In the figure we can see the effect of the threshold
organize them in a [u] matrix, which is part of the file selection in the overall efficiency of the system until
system. certain value is reached, from which on, the success rate
Each vector in matrix A is also projected to the remains constant
banknotes space, using eqn (12), where [u]T is the transpose
of the eigenvectors matrix, which is also the change of
basis matrix, introduced in the PCA theory. As a result, a
matrix [Ω] is produced, which contains, by columns, each
one of the sample images from the training set, but
represented in the new space, as 24×1 vectors, called
samples weights vectors.

[Ω]=[𝑢]𝑇 A (12)

Those vectors will be compared one by one, within the

real- time system, with an input weights vector.
FIG. I.B.5 The Output Graph
3) Representation of Input Image into New Space

If, as performed for all sample images of the training

set, a 60×90 input image in the real-time system from Fig. III. RECOGNITION OF COINS
7 is represented as a 5400×1 vector (Γi ), it can also be
represented as a 24×1 vector called input weights The Recognition of Indian coins may be familiarized
vector:[Ωin ]=[ω1 ω2 … ωK ], by using Ψ and each eigen since it has got various methods to implement it out like the
image ui , both loaded from the file system through the ones in coin counting machines which just checks out the
projection defined in eqn (13) which obtains each 𝜔𝑖 size and weight of the coin. But this method is not apt
component of the input weights vector. because there may be coins of different denomination of
the same size which makes the recognition part difficult.
Hence we come up with this method constituting the
𝜔𝑖= 𝑢𝑖 𝑇 (𝛤𝑖 - 𝛹),∀ 𝑖 =1… K (13) Rotation Invariance method and Image Subtraction.

4) Mahalanobis Distance Of the Image

A) Rotation Invariance Method
To determine the class to which the input image
belongs, we use the Mahalanobis distance dm between Ωin
and each one of the Ωi samples weights vectors from the
training set in [Ω], by using eqn(14), where λj is each one of
the K=24 eigenvalues of the original covariance matrix in
the file system’s vector rather than a variance, because
each eigenvalue is proportional to a corresponding variance
of the original covariance matrix
(Ω𝑖𝑛 𝑗 −Ω𝑖 𝑗 )
Mahalanobis distance 𝑑𝑚 = ∑𝐾
𝑗=1 ,
𝜆𝑗
∀ 𝑖 =1…K

5) BankNote Recognition

An image corresponding to the input weights vector is

classified as a particular denomination banknote (and this
denomination outputted as an audio message) if the
minimum distance from Ωin to Ωi each projection from the
training set is smaller than a certain threshold; otherwise,
the image is discarded and no output message is emitted.
The threshold was experimentally determined, from a set of
test images corresponding both to banknotes of several
denominations and random images. The results of this
experimentation are shown in Fig.I.B.5, where the success
rate in the process is compared against the threshold value

Volume 3, Issue 05 Published by, www.ijert.org 4

Special Issue - 2015 International Journal of Engineering Research & Technology (IJERT)
ISSN: 2278-0181
NCETET-2015 Conference Proceedings

The proposed approach of coin recognition consists of

five modules namely, image acquisition, image SUBTRACTED (r,c) = OBJECT(r,c) – TEST(r,c)
segmentation, radius calculation, image subtraction and (14)
Threshold comparison. Input image of size (320*320
pixels) is acquired and coin is segmented through it.
Fig. 2.1 elucidates the block diagram of the proposed
methodology.

1) Image Procurement and Detailing

This section describes the way in which the images are

procured and the way by which they are segmented and
hence suitably being brought down to an level where further FIG III.A.2 Subtraction of Coins
phase of recognition techniques can be done so that the
coin can be recognized well enough. First it is converted to b) Fine Subtraction: Here in this subtraction which
its grayscale and then it is further adjusted by increasing its is same as that of the course subtraction here we reduce the
contrast and getting its binary form. Each of these are step size from 10º to 1º so that the calculation be more
described in the FIG.III.A.1. The binary image is finite than the course.
transversed row wise and we get the end of diameter.Thus
we get the exact position of the image. Hence when both these subtraction are done we get a
resultant value for both the subtraction. We get the minima
values which give more dark patches to the out image we
get.

4) Threshold Comparison and Coin Recognition

Once we get the minima of the gray value sum,

based on comparison with a standard threshold,
deductions are made whether the coin matches or not.
If the minimum value lies below the threshold, coin
identification is established difference between the
object and test image.
FIG III.A.1 The Prescribed Coin Formats
The identified coin denomination is heard out as an
audio into the visually impaired ears through the
earphone.
2) Calculating the Radius
IV. THE RESULT
Diameter is calculated by finding the difference
between maximum and minimum position of white The proposed system for the Indian currency recognition
pixels of the binary image formed during image has been implemented in MATLAB software. The result for
segmentation shown in the above Fig. this cardinal step the recognition of both notes and coin has been
provides the value based on which the suitable image implemented down accordingly. The bank notes recognition
from the database gets selected for further process, by the mahalanobis distance and the image subtraction
abridging the process time and irrelevant data since the technique for the coin has been implemented well off.
Indian coins have distinct radius. The results obtained from both these substantial
methods has been folded up and shown in various figurative
3) Rotation Invariance Image Subtraction ways. It has been verified and set out in this paper for its
research senses.
After we have procured both the object and the test A success is considered if, by an appropriate use
image. We get into the main constituent of coin recognition methodology, the system has emitted an audio message
called the Image Subtraction by Rotation Invariance corresponding to the denomination of the banknotes or the
method. The recognition part consists of two main checks to coin in test, at most, at the second try; no result, if, after
narrow down the mistakes or to make it accurate.
those two attempts, no message has been produced. A false
a) Course Subtraction: In this subtraction the test
positive is considered if the system emits a denomination
image is given one full rotation in steps of fixed angular
distance of say 10º. At each instance of rotation image different than that of the object in test. The issuance of a
subtraction is carried between the rotated test image and false positive can be avoided by working in near-ideal
the input object image. conditions.

Volume 3, Issue 05 Published by, www.ijert.org 5

Special Issue - 2015 International Journal of Engineering Research & Technology (IJERT)
ISSN: 2278-0181
NCETET-2015 Conference Proceedings

V. CONCLUSION We like to thank our Principal Fr.JoseKanampuzha for

A portable and easy access system for image recognition giving us his permission for all our activities we needed to
can be implemented using the technology of cameras and do and also for his well wishes
the theory of the Digital Image Processing.
One of the main constraints of the system developed in We give thanks to our H.O.D Prof.Satheesh Kumar for
this paper is the fact that the background of the image giving us the support in doing this paper, and also a big
containing the object to identify (i.e. the currency), must be thanks to our mentor Prof.Anu Abraham Mathew for his
contrasting with that object. Another constraint is that the valuable advice and his constant inspiration for all that we
illumination conditions over the image must be uniform. did.
This solves a day to day problem which takes place in
the life of the visually impaired people. REFERENCE
Future works will include modifications of the technique
[1]. Felipe Grijalva, J.C. Rodríguez, Julio Larco and Luis Orozco ,”
and also merging of other image processing techniques, Smartphone Recognition of the U.S. Banknotes’ Denomination, for
such as,Neural Networks training using Edge detection Visually Impaired People” ©2010 IEEE
which wouldextricate the process from the dependency over [2]. M. Turk & A. Pentland, "Face recognition using eigen faces,“ J.
standard light intensity and standard distance between Cognitive Neuroscience, vol. 3, no. 1, pp. 71–86, 1991.
[3]. R. Gonzalez and R. Woods, “Digital Image Processing”, 2nd New
image and camera duringimage acquisition adding on to the Jersey, U.S.A.: Prentice Hall, 2002.
accuracy of the process. [5] E. Ashbridge, D.I. Perrett, M.W. Oram and T. Jellema, (2000)“Effect
of Image Orientation and Size on Object Recognition: Responses of
ACKNOWLEDGMENT Single Units in the Macaque Monkey Temporal Cortex”, Cognitive
Neuropsychology Vol. 17: 1/2/3, pp. 13–34
[6] R. Bremananth, B. Balaji, M. Sankari, A. Chitra, “A New Approach to
We would like to thank our institution AmalJyothi Coin Recognition Using Neural Pattern Analysis”, 2005 annual
College of Engineering for giving us space and time to do IEEE,Indicon, pp. 366-370
the research and findings for this paper.

Volume 3, Issue 05 Published by, www.ijert.org 6