International Journal of Computer Science Trends and Technology (IJCST) – Volume 11 Issue 3, May-Jun 2023

RESEARCH ARTICLE OPEN ACCESS

Currency Exchange System Using Image Processing

V Ramya [1], A Sai Deepika [2], K Rupa [3], M Leela Krishna [4],
I Satya Girish [5]
[1]
Asst. Professor, Department of Computer Science and Engineering
[2],[3],[4],[5]
Student, Department of Computer Science and Engineering
[1], [2], [3], [4], [5]
QIS College of Engineering and Technology - Ongole

ABSTRACT
In this paper, we proposed an automatic currency recognition system using digital image processing methodology.
The following project mainly focuses on the recognition of currency by its image or photograph. It will help users to
recognize details about currency like currency value, currency name, the value in inr, euro and us dollar. It works
using the main characteristics of currency notes such as size colour or printed text on it and also depends on differ in
currency notes within the same country. We have considered indian rupee and us dollar, the most used currencies in
our domain with their denominations. This system works accurately and also able to quickly identify the currency
notes.

I. INTRODUCTION
According to the un charter there are around 195 same as the expected output. First of all input a
countries around the globe. In which 193 countries currency note image, which needs to be checked.
are members of the un and two are observing states. Then system performs basic image processing
According to the u.n., worldwide there are 180 techniques on input image and refined to convert it
currencies. All these currencies are different in into a usable input for matching with present image
characteristics such as size, colour and texture. In the in dataset. The system extracts the information from
era of rapidly growing levels of trade between image based on features such as colour and texture.
countries and also tourism all over the world, it With the help of these features the system determines
becomes necessary to recognize each currency note currency name and denomination of the currency
correctly. Now a days people travel to different note. After this the exchange rate of currency will be
countries, they use their native country currency in extracted from internet with the help of online
paying bills or buying stuffs and because most of the exchange rate api url. At the end output displayed on
local people are not familiar with the currency other the screen.
than their own country currency and also not familiar
with the exchange rate of that currency in their own II. LITERATURE SURVEY
currency, it becomes necessary to develop an Digital image processing, 4th edition
automated system that helps in recognition notes The 4th edition, which celebrates the book’s 40th
easily, faster with efficiency. anniversary, is based on an extensive survey of
The proposed system is based on image processing faculty, students, and independent readers in 150
and makes the process robust and automatic. institutions from 30 countries. Their feedback led to
We used inr and usd as an example to illustrate the expanded or new coverage of topics such as deep
technique. This system is based on our knowledge learning and deep neural networks, including
about computer science technologies like digital convolutional neural nets, the scale-invariant feature
image processing, python and also a small step to transform (sift), maximally-stable extremal regions
implement in a system that is most important for (msers), graph cuts, k-means clustering and
industrial development. superpixels, active contours (snakes and level sets),
We used inr and usd as an example to illustrate the and exact histogram matching. Major improvements
technique. This system is based on our knowledge were made in reorganizing the material on image
about computer science technologies like digital transforms into a more cohesive presentation, and in
image processing, python and also a small step to the discussion of spatial kernels and spatial filtering.
implement in a system that is most important for Major revisions and additions were made to examples
industrial development. and homework exercises throughout the book. For
The block diagram as in fig. 1 shows the used the first time, we added matlab projects at the end of
methods in the system to obtain the actual output

ISSN: 2347-8578 www.ijcstjournal.org Page 12

International Journal of Computer Science Trends and Technology (IJCST) – Volume 11 Issue 3, May-Jun 2023

every chapter, and compiled support packages for contains rmb serial number. Then the detected text
students and faculty containing, solutions, image region is binarized by a combined thresholding
databases, and sample code. technique. After that, a local contrast average method
Identification of paper currency techniques: a survey is introduced to extract the rmb characters from the
Paper currency identification is an image processing binarization result. The experiments demonstrate that
technique i.e. Worn to recognize currency of different the proposed binarization method outperforms other
countries. The paper currencies of different countries well-known methods. For character extraction, we
are possibly interweaved collectively consequently report an overlap-recall rate of 79.68% and an
rises ever more. It is a challenge for standard overlap-precision rate of 98.10% respectively.
currency recognition systems. However, the main Digital image processing and analysis: human and
focus of most of the standard currency recognition computer vision applications with cviptools
systems and machines is on recognizing forged The author presents the book on digital image and
currencies. Hence there is very vital role of currency analysis that has four sections and thirteen chapters,
identification system and it is essential that the which is written at a junior-year or above level and
identification system should be very accurate. A used as a basis for advanced studies involving
thriving approach for paper currency identification images. The first section deals with introduction and
depends upon preprocessing, feature extraction and overview of digital image processes and analysis
classification of that currency image. In this paper, (chapter 1), and computer imaging systems (chapter
we have gone through different literature which 2). Section 2 contains digital image analysis and
describes different techniques of paper currency computer vision that includes chapters: 3
recognition. Finally we have concluded that if we (introduction of digital image analysis); 4- image
apply some efficient pre-processing and feature segmentation and line and edge analysis; 5- discrete
extraction technique than we can improve the transform (fourier and other), and filtering; and 6-
accuracy of identification system. feature analysis and pattern classification. Section 3
Recognition system for real time paper currency covers digital image processing and human vision,
Money number recognition refers to the money of the and consists of chapters: 7- human vision and
currency, the currency and authenticity recognition. perception and heavily relates to the usefulness of
Money number recognition system is the kernel image processing and the effects of noise reduction;
module of self-service system, and the major applied 8- overview of image enhancement techniques; 9-
range is cash-related equipments. In this paper we image restoration; and 10- image compression with
design a kind of money number recognition system. discussion of entropy and information. And section 4
The quick positioning of money number is achieved deals with program and application development with
based on gray value accumulation. The edge line of cviptools (computer vision and image processing
money number area is detected using the least square (cvip)), and has the following chapters: 11- basic use
method. Using geometrical rotation method and gray of the cd software; 12- application development; and
adjacent interpolation method to realize the number 13- libraries.
of tilt correction. Based on the character structure An intelligent system for paper currency recognition
characteristic and the imaginary line and character of - robust features
the point of intersection features, formation Intelligent systems on paper currency recognition and
recognition judgment tree, realized the character verification are inevitable for modern banking
recognition. The simulation experiment indicates that services. These systems are used in auto-seller
this algorithm has a high recognition accuracy under machines, vending machines etc. Extracting
the condition of rejection. sufficient and reliable monetary characteristics are
Extraction of serial number on bank notes essential for accuracy and performance of such
The study of rmb (renminbi bank note, the paper systems. This paper proposes a new intelligent
currency used in china) serial number recognition system for paper currency recognition. Pakistani
draws more and more attention in recent years, for paper currency has been considered, as a case study,
reducing financial crime, improving financial market for intelligent recognition. This paper identifies,
stability and social security. The accuracy of rmb introduces, and extracts robust features from
recognition relies heavily on the extraction, which is pakistani banknotes. After extracting these features,
a challenging problem due to background variations the paper proposes to use three layers feed-forward
and uneven illumination. In this paper, we present a backpropagation neural network (bpn) for intelligent
new system that extracts the rmb characters directly classification. The proposed technique and system are
from scanned rmb images. First, two different simple and comparatively less time consuming which
techniques, namely skew correction and orientation makes it suitable for real-time applications. In order
identification are used to detect the region which to evaluate the performance of the proposed

ISSN: 2347-8578 www.ijcstjournal.org Page 13

International Journal of Computer Science Trends and Technology (IJCST) – Volume 11 Issue 3, May-Jun 2023

technique, experiments have been conducted on 175 characteristics such as size, colour and texture. In the
pakistani banknotes. The results indicate that system era of rapidly growing levels of trade between
has 100% recognition ability on properly captured countries and also tourism all over the world, it
images. becomes necessary to recognize each currency note
Paper currency recognition - journal of advanced correctly. Now a days people travel to different
computer science and technology research countries, they use their native country currency in
Paper currency recognition (pcr) is an important area paying bills or buying stuffs and because most of the
of pattern recognition. A system for the recognition local people are not familiar with the currency other
of paper currency is one kind of intelligent system than their own country currency and also not familiar
which is a very important need of the current with the exchange rate of that currency in their own
automation systems in the modern world of today. It currency, it becomes necessary to develop an
has various potential applications including electronic automated system that helps in recognition notes
banking, currency monitoring systems, money easily, faster with efficiency. Local binary patterns
exchange machines, etc. This paper proposes an (lbp) is a type of visual descriptor used for
automatic paper currency recognition system for classification in computer vision and is a simple yet
paper currency. A method of recognizing paper very efficient texture operator which labels the pixels
currencies has been introduced. This is based on of an image by thresholding the neighborhood of
interesting features and correlation between images. each pixel and considers the result as a binary
It uses radial basis function network for number. Due to its discriminative power and
classification. The method uses the case of saudi computational simplicity, lbp texture operator has
arabian paper currency as a model. The method is become a popular approach in various applications. It
quite reasonable in terms of accuracy. The system can be seen as a unifying approach to the traditionally
deals with 110 images, 10 of which are tilted with an divergent statistical and structural models of texture
angle less than 15o. The rest of the currency images analysis. Perhaps the most important property of the
consist of mixed including noisy and normal images lbp operator in real-world applications is its
50 each. It uses fourth series (1984–2007) of robustness to monotonic gray-scale changes caused,
currency issued by saudi arabian monetary agency for example, by illumination variations. Another
(sama) as a model currency under consideration. The important property is its computational simplicity,
system produces accuracy of recognition as 95.37%, which makes it possible to analyze images in
91.65%, and 87.5%, for the normal non-tilted images, challenging real-time settings.
noisy non-tilted images, and tilted images Disadvantages
respectively. The overall average recognition rate for • Very critical to recogition
the data of 110 images is computed as 91.51%. The • Time taken process
proposed algorithm is fully automatic and requires no
human intervention. The proposed technique IV. PROPOSED SYSTEM
produces quite satisfactory results in terms of The proposed system is based on image processing
recognition and efficiency. and makes the process robust and automatic.
Study of printing identification based on We used inr and usd as an example to illustrate the
multispectrum imaging analysis, proceedings of the technique. This system is based on our knowledge
international conference on computer science. about computer science technologies like digital
This paper introduced and adopted multispectrum image processing, python and also a small step to
imaging analysis to conduct the corresponding implement in a system that is most important for
research and the exploration for the purpose of industrial development. We had considered indian
effectively solving the problems of printing rupee and us dollar for this project. Project can
identification such as credentials, bills, legal expand for more currencies inclusion according to
documents and so on. In addition, it designed and had use.
realized a method to identify printing that had laid a Algorithms used
foundation for its more extensive application in Edges are significant local changes of intensity in a
printing identification field. digital image. An edge can be defined as a set of
connected pixels that forms a boundary between two
III. EXISTING SYSTEM disjoint regions. There are three types of edges:
According to the un charter there are around 195 Horizontal edges
countries around the globe. In which 193 countries Vertical edges
are members of the un and two are observing states. Diagonal edges
According to the u.n., worldwide there are 180 Feature extraction
currencies. All these currencies are different in

ISSN: 2347-8578 www.ijcstjournal.org Page 14

International Journal of Computer Science Trends and Technology (IJCST) – Volume 11 Issue 3, May-Jun 2023

Edge detection allows users to observe the features In conclusion, we have designed a system that
of an image for a significant change in the gray accurately identifies both the country of origin and
level. This texture indicating the end of one region the denomination of a given banknote. Our system
in the image and the beginning of another. It currently supports twenty of the most common
reduces the amount of data in an image and currencies, but can easily be extended to more
preserves the structural properties of an image. countries based on the method we have previously
Cnn algorithm described. When compared with the crude algorithm
1) feature extraction: cnn compose of multiple of pixel by pixel comparison, our algorithm is
layers and first layer define for feature extraction and considerably more accurate, and takes less time. We
this features will be extracted from given input image have thus learned that our proposed algorithm is able
dataset or any other multidimensional dataset. to identify currency and denomination in an average
2) feature selection: using this layer features of 5.3 seconds, which is a considerable improvement
will be selected by applying a layer called pooling or over the crude algorithm. However, our proposed
max polling. system only considers a limited number of
3) activation module: using this module relu currencies. There are 180+ currencies that can be
will be applied on input features to remove out included in the system, and we have chosen to only
unimportant features and hold only relevant do for 20 of the most common ones. Also, the system
important features should be effective in identifying notes that are
4) flatten: this layer will be define to convert mutilated. Our system is not effective under this
multidimensional input features into single consideration. This can be worked on in the future.
dimensional input array
5) dense: this layer can be used to connect one REFERENCES
layer to other layer to receive input features from 1. Pearson book on digital image processing 4th
previous layer to new layer to further filter input edition by gonzalez (2018).
features in next layer to get most important features 2. Programming and problem solving with python
from dataset to have best prediction result. 1st edition by ‘ashok namdev kamthane’ and ‘amit
Advatages: ashok kamthane’ 2017
• By using machine learning currency 3. Identification of paper currency techniques: a
recognition is very easy survey by s. Sahu and t. Verma international journal
• Very less time for currency recognition. of science technology & engineering (2016)
4. Recognition system for real time paper currency.
V. IMPLEMENTATION Ieee 2012.
Open image: 5. Extraction of serial number on bank notes. Ieee.
When you click on this button then it will direct you 6. Digital image processing and analysis: human and
to memory and ask you to select image of currency. computer vision applications with cviptools, 2nd
After selecting an image you will automatically edition by scott e umbaugh, t and f india.
return to the window and your selected image will be 7. An intelligent system for paper currency
visible in application window. recognition - robust features journal of intelligent &
Recognize: fuzzy systems - a. Sargano, m. Sarfrazb, and n. Haq,
This button is used for currency recognition. After 2014.
pressing this button currency recognition will start 8. Paper currency recognition - journal of advanced
and you will get output in few seconds on the computer science and technology research , ‘althafiri
application window. e, sarfraz m’, ‘alfarras m. Bahraini’.
Conversion 9. Study of printing identification based on
This button is used for currency conversion. After multispectrum imaging analysis, proceedings of the
pressing this button currency conversion will start international conference on computer science and
and you will get output in few seconds on the software engineering, by liu q, tang l. P. 229 – 232;
application window. 2008.
Reset: 10. Https://www.googlescholar.com
this button is used to clear the application window. 11. Https://www.geeksforgeeks.org/digital-
Exit: imageprocessing/
this works as simple exit button, you will come out 12. Https://www.elprocus.com
of the application and all processes of application 13. Https://www.worldometers.info/geography/
will be terminated. 14. Https://www.worldatlas.com/articles/how-
manycurrencies-are-in-the-world.html
VI. CONCLUSION

ISSN: 2347-8578 www.ijcstjournal.org Page 15