Detection of Fake Currency Notes using Image

2023 Annual International Conference on Emerging Research Areas: International Conference on Intelligent Systems (AICERA/ICIS) | 979-8-3503-0345-2/23/$31.00 ©2023 IEEE | DOI: 10.1109/AICERA/ICIS59538.2023.10420367

Processing Technique
Nagendra Kumar Naseem Ahmad Amit Mittal
Department of Elect. & Elect. Engg. Department of CSE Department of Allied Sciences
GL Bajaj Institute of Tech. & Manag. Dev Bhoomi Uttarakhand University Graphic Era Hill University
Greater Noida, 201306, India Dehradun, India Bhimtal 263136, India
nagendra.k962@gmail.com naseem.saifi2011@gmail.com amitforestry26@gmail.com

Akhilesh Singh Sanjay Oli Ashish Rayal

Department of Electrical Engg. Department. of Mathematics Department of Mathematics
Uttarakhand Technical University Dayananda Sagar College of Engg. Uttaranchal University
Pithoragarh Campus (NPSEI), 262501 Bangalore 560078, India Dehradun-248007, India
akhil09msingh@gmail.com sanjay-maths@dayanandasagar.edu ashishrayal@uttaranchaluniversity.ac.in

Abstract— One issue that comes up in financial transactions Every note in transactions must be manually tested, which
is substitute notes. for nations like India, the United States, and is a time-consuming and messy operation. There is also a
several more in Europe. That is attractive a significant barrier. potential that notes will rip while being handled. In many
The technology that aid in scanning and printing. With the use applications, including automatic selling systems and vending
of modern equipment, a group of individuals can print and copy machines, automatic methods for bank note identification are
notes. Because physically identifying fraudulent cash is a dirty therefore necessary. For the automated system to be accurate
and time-consuming operation, there is a need for automated and reliable, the currency picture must be sufficiently
techniques and software that can quickly identify false money. extracted for monetary properties.
Here, we present the concept in two straightforward ways: one
by analyzing images captured in small spectral bands across a A. Method of Edge detection
larger band Create spectra over the whole spectral range for
Edge detection is a method that is used in digital image
each and every pixel in the picture. And the last way is the
processing to identify and extract the boundaries or edges that
Extraction of many additional elements in both actual and We
can distinguish between authentic and phony cash notes by
separate various objects in an image. A comparison of the
placing duplicate notes side by side. In our suggested effort, we photo in question to an appropriate reference image may
mix these two and structure the steps forward. The different achieve this goal. The edges of an image give vital information
color lights employed by the are Ultraviolet (UV), light hyper about changes in the picture's hue or brightness in computer
spectrum imaging, normal light emitting diode, green light vision. Edge detection is the process of applying an algorithm
emitting diode and blue light emitting diode, and red light to a photograph in order to discover areas where the image's
emitting diode, with diverse wavelengths ranging from 350 nm color or brightness fluctuates considerably. These areas are
to 805 nm, corresponding. therefore referred to as borders or edges. This method is used
to analyse a photograph in order to identify regions with
Keywords— MATLAB2019a, entropy, HIS (hyper spectral noteworthy changes in brightness or hue. This approach is
imaging) used to find parts of a picture that have noticeable differences
in brightness or colour. This is performed by highlighting and
I. INTRODUCTION identifying certain portions of the image that best depict these
The country's sole power to issue notes is the Reserve variances. The method may be used to detect areas in an image
Bank. We are in charge of designing, producing, and that exhibit significant fluctuations in either of these
managing the nation's currency overall in conjunction with the properties by locating spots that demonstrate both color and
Government of India to guarantee a sufficient supply. of pure brightness changes at the same time. The goal of this
and sincere notes Like other central banks across the world, procedure is to find portions of the image that have
Reserve Bank occasionally modifies the appearance of bank noteworthy variations in brightness or color. The
notes. Traditionally, this has required anti-counterfeiting implementation of this technology is often referred to as "edge
techniques, such as enhancing banknotes with fine intaglio detection." Edge detection filters often used include the
detail to make it simple for laypeople to see fake money. To Canny, Laplacian, and Sobel filters. Edge detection is a
demonstrate that no precious metal has been scraped off of critical component in the science of computer vision since it
coins, they may be milled or marked with parallel grooved has so many practical applications. These applications include
edges. The Reserve Bank employs a number of methods to image segmentation, object tracking, and object recognition,
identify counterfeit money [1]–[5]. In [1], the authors used among others. Another sector that employs this approach is
HOG encoding to extract features from the banknote's latent artificial intelligence (AI), which also utilizes a technology
picture. K-means and SVM classifiers have been used to falling within this classification. The use of this technology
categories fraudulent Indian banknotes. In [3] and [6], the enhances the process of extracting crucial information from
authors employed pattern matching to verify the validity of pictures, hence improving the precision of image analysis and
Indian rupee notes while also employing color-based expanding the volume of knowledge that can be acquired.
denomination recognition. K-NN Technique is used [7] while
Minimum Distance Classifier Technique [8] are used to detect B. Method of Image segmentation
Fake Currency. Soft computing techniques are applied to The phenomenon of dividing an image into many
tackle different real life problems such as prediction issues [9], segments or sections, where each segment corresponds to a
[10], [11], decision issues [12], [13], control issues [14]–[17], unique part or component of an object, is often known as
reliability analysis [18], [19] and detection of fake currencies picture segmentation. This method has the potential to be
[20]–[23] and etc. executed either manually or mechanically. Image

979-8-3503-0345-2/23/$31.00 ©2023 IEEE

Authorized licensed use limited to: Jeju Nat'l University. Downloaded on March 01,2025 at 09:33:20 UTC from IEEE Xplore. Restrictions apply.
segmentation is the process of splitting a picture into different features, such as watermarks, holograms, and micro printing,
pieces. Image segmentation is mainly used to minimize an to make the notes even more secure.
image's complexity while increasing visual clarity and
comprehensibility, allowing for more effective assessment B. Serial number
and analysis. This may be performed by modifying or With rising ticket numbers from tiny to big in the upper
simplifying the original image. Segmentation is the process of left and lower right corners, the serial number panel.
giving a label or value to each pixel in a picture. This label or
C. Latent image
value is determined by the characteristics of the pixel in
question as well as the pixels that are immediately around it. A latent image is an invisible, temporary image that is
Segmentation is accomplished by using a computer. This created on a photosensitive surface, such as photographic film
group of qualities could include things like color, texture, and or a digital camera sensor, before it is developed or processed.
intensity, in addition to perhaps include other visual elements. D. Watermark
The domains of computer vision and image processing both
make substantial use of a method known as picture Likeness of Mahatma Gandhi, multiway lines and numeri
segmentation. This technique is used for a broad variety of cal denomination mark appear visible in light.
applications, including object detection and tracking, as well E. Identification mark
as picture compression and medical imaging. It allows
A latent image is an invisible, temporary image that is
computers to recognize and analyze specific objects within an
created on a photosensitive surface, such as photographic film
image, enabling them to perform tasks such as counting
or a digital camera sensor, before it is developed or processed.
objects, measuring their size and shape, and detecting
abnormalities. The process to be followed is given in flow chart.
II. METHODOLOGY OF PROPOSED WORK Step 1: Image acquisition:
This tactic work on the figure or images of banknotes real Image acquisition is the process of capturing digital
time develop by a digicam is setup and capture the image of images of real-world scenes or objects using an imaging
the various currency or upload the images of currency device, such as a camera, scanner, or sensor. The method
specially notes which was save in the folder. The requires converting the optical information of the scene or
breakthrough which is used in this case is as follows process item being analyzed into a digital format that is capable of
shown in Fig. 1: being processed by a computer system. This makes it possible
1. With a basic digital scanner or camera that was attached for the information to be read by the computer and processed
to the system, it was possible to capture a picture of a money by it shown in Fig 2.
note under normal lighting.
2. The outcome is an RGB image that has been changed
to a grayscale image.
Image Acquisition
3. Detecting edges in photos in grayscale.
4. Pre-Processing - A reliable strategy for noise reduction, Gray Scale Conversation
the median filter is frequently
utilized. This filter works very well to reduce "salt-and-
pepper" noise.
5.Now the banknote features will be cut and segmented. Edge Detection
6. After segmentation, the features are extracted from the ba
nknote
7. After Extraction Region of Interest (ROI) will be Image Segmentation
calculated
8. Selecting the ROI figure and then system perform
various technique to identify the note is real or fake Characteristics
7. If all the condition is satisfied then new Window is Extraction
open and pop up the notification whether the bank is fake or
original
This method uses the characteristics of the currency used
by ordinary people to distinguish between different denomin Extraction Of ROI
ations of banknotes. Features that can be used to verify the a
uthenticity of banknotes are
A. Security Filament Result
The security filament in Indian currency notes has the
denomination value printed on it in numerals. The color of the Fig. 1. Flow chart of the Process.
filament also varies according to the denomination of the note.
For example, the security filament in the 2000-rupee note is
pink, while that in the 500-rupee note is blue. This feature
makes it difficult for counterfeiters to replicate as it requires
specialized printing technology to produce. Additionally, the
security filament is often combined with other security

Authorized licensed use limited to: Jeju Nat'l University. Downloaded on March 01,2025 at 09:33:20 UTC from IEEE Xplore. Restrictions apply.
 Step 3: Edge Detection
The technique of determining the limits or edges of
features or objects included inside an image is referred to as
edge detection. Object identification, segmentation, and
tracking are three examples of the numerous image processing
and computer vision applications that rely on this phase as a
key building block. In edge detection, the algorithm examines
the variations in intensity or color between neighboring pixels
in order to discover abrupt changes or discontinuities in the
picture. It's possible that a number of reasons are to blame for
these disparities. The presence of an object border, a texture
or pattern, a change in shade or lighting, or any combination
of these elements may be responsible for bringing about these
adjustments shown in Fig. 4.

Fig. 2. Acquired image of 500 and 2000 notes.

Step 2: Gray scale conversion

The process of transforming a colored picture into a
grayscale or black-and-white image is referred to as grayscale
conversion, shown in Fig. 3. The pixels in a grayscale picture
each have a single value that indicates the brightness or
intensity of the pixel, but there is no information about the
color of the pixel. This sort of digital image is called a
grayscale picture, and it is one kind of digital image. Each
pixel in a color picture has a mixture of red, green, and blue
color channels, which come together to form the pixel's
overall color. The values of the color channels are mixed in
grayscale conversion in such a manner that the luminance or
brightness of the pixel as a whole is maintained, but the color
information is lost.
Fig. 4. Edge Detection of 500 and 2000 notes respectively.

Step 4: Image Segmentation

A digital picture may be segmented via a technique known
as image segmentation. This process involves the division of
the digital image into many segments or regions, each of
which corresponds to a separate item or area of interest in the
image. picture segmentation seeks to achieve, as its primary
objective, the simplification or transformation of the
representation of a picture into something that is both more
meaningful and simpler to evaluate. When an image is
segmented, individual pixels or groups of pixels that have
similar qualities, such as color, intensity, texture, or motion,
are grouped together into segments or regions. These
segments or regions may then be analyzed individually.
Different degrees of granularity may be achieved while
carrying out segmentation, ranging from the identification of
individual items to the segmentation of areas included inside
those objects.
Step 5: Feature Extraction
In the context of banknotes and money, the process of
Fig. 3. RGB to Gray Scale Conversation of 500 and 2000 notes extracting certain characteristics or patterns that are unique to
respectively. a particular currency or banknote and that may be used for

Authorized licensed use limited to: Jeju Nat'l University. Downloaded on March 01,2025 at 09:33:20 UTC from IEEE Xplore. Restrictions apply.
authentication or verification is referred to as feature 2 The distribution of colors or values indicating the
extraction. Feature extraction can be used for either strength of color in various areas of the banknote
authentication or verification reasons. A wide variety of
security features, such as watermarks, security threads, 3 The spatial frequency characteristics of the many
holograms, micro printing, and other printing processes that different patterns or features that are printed on the
are difficult to copy may be included in these features. The banknote
procedure of feature extraction may be carried out using a 4 Analysis of the different textures present on the
wide variety of strategies, such as algorithms for image different parts of the banknote
processing and procedures for machine learning, amongst a
great many other options. The process of feature modeling is 5 Specific geometric elements on the banknote, such as
another name for feature extraction. The pictures of the corner points, edges, or contour curves. These may be
banknotes are analyzed using these approaches, and the found on the banknote.
particular characteristics and patterns that are necessary for
authenticity or verification are extracted shown in Fig. 5-8.

Fig. 5. Edge based segmentation of mahatma Gandhi 500 and 2000 notes Fig. 7. ROI Extraction texture & Statistical Features of 500 and 2000
respectively. notes respectively.

Specific qualities of the edges or borders of different

patterns or features on a banknote are referred to as the edge
features of the banknote. Edge features are also known as
border features. These characteristics are often retrieved by
image processing algorithms or approaches that include
machine learning and are utilized for reasons related to
authentication or verification.
The contour and curvature of certain edges or borders on
the banknote, such as the edges of a watermark or a security
thread, are examples of edge characteristics that may be
retrieved.
Fig. 6. Identification mark
1 The shape and curvature of specific edges or borders
Statistical Features The quantitative measurements or on the banknote.
statistical characteristics of numerous features or patterns on a 2 The frequency or density of edges or borders in
banknote are referred to as the statistical features of banknotes. specific regions of the banknote, such as the edges of
Statistical features are also known as security features. It is fine line patterns or micro printing.
standard practice to get these qualities by making use of
algorithms for image processing or approaches that are based 3 The sharpness or smoothness of edges or borders in
on machine learning when one is authenticating a person or specific regions of the banknote, which can indicate
validating their credentials. This may be done in a number of the quality of the printing or manufacturing process.
different ways. 4 The color or intensity changes along the edges or
1 The length, width, and thickness of the banknote, in borders of specific features on the banknote, such
addition to the note's aspect ratio. as the edges of a portrait or emblem.

Authorized licensed use limited to: Jeju Nat'l University. Downloaded on March 01,2025 at 09:33:20 UTC from IEEE Xplore. Restrictions apply.
 instances into one of two groups. As using money is essential
to surviving, it is always important to monitor its uniqueness.
In India, paper money is used considerably more often,
necessitating the development of a mechanism to identify
counterfeit money. The suggested approach appears to be
helpful in determining if the new currencies being utilised in
the market are real or not. In addition, rather than just
presenting the outcome, it indicates where the currency
discrepancies are. This experiment demonstrates that the
Indian banknote may be detected using a low-cost device. and
the outcome is 95.0%, demonstrating the effectiveness of the
approach.
III. CONCLUSION
As using silver is essential to surviving, it is always
important to adhere to its original design. India uses a lot of
banknotes; hence a mechanism is required to spot fake
currency. The suggested approach appears to be helpful in
determining if the new currencies being utilised in the market
are real or not. In addition, rather than just presenting the
outcome, it indicates where the currency discrepancies are.
This experiment demonstrates that the Indian banknote may
be detected using a low-cost device. and the outcome is
95.0%, demonstrating the effectiveness of the approach.
Fig. 8. ROI-Extract Edge & Shape Feature.
IV. FUTURE SCOPE
Step 6: Result Authors would like to develop an Android app in the future
to identify Indian money. Future applications of this system
The result is shown in MATLAB which show the window
might include international currencies like dollars, euros, taka,
that currency is fake or original. (See Fig. 9)
etc.
REFERENCES
[1] M. Singh, P. Ozarde, and K. Abhiram, “Image Processing Based
Detection of Counterfeit Indian Bank Notes,” in 2018 9th
International Conference on Computing, Communication and
Networking Technologies (ICCCNT), Jul. 2018, pp. 1–5. doi:
10.1109/ICCCNT.2018.8493763.
[2] D. Kumar and S. S. Chauhan, “a Study on Indian Fake Currency,” Int.
J. Creat. Res. Thoughts, vol. 8, no. 3, pp. 568–573, 2020.
[3] M. P. Deepak and N. B. Prajwala, “Identification of Fake Notes and
Denomination Recognition,” in 2018 International Conference on
Communication and Signal Processing (ICCSP), Apr. 2018, pp.
0238–0243. doi: 10.1109/ICCSP.2018.8524275.
[4] K. Chinthana, K. N. Mahendra, and Y. S. Sahana, “Feature Extraction
and Identification of Indian Currency for Visuallu Impaired People,”
pp. 1–3, 2018.
[5] I. A. Suresh, “Indian Currency Recognition and Verification Using
Image Processing,” 2016.
[6] K. B. Zende, V. Kokare, S. Pise, and P. S. Togrikar, “Fake Note
Detection System,” Int. J. Innov. Res. Technol., vol. 4, no. 1, pp. 46–
49, 2017.
[7] Y. Neeraja, B. Divija, and M. Nithish Kumar, “Fake Currency
Detection Using K-NN Technique,” Int. J. Res. Eng. IT Soc. Sci., vol.
565, no. 1, pp. 201–205, 2019, [Online]. Available:
http://indusedu.org
[8] K. Sawant and C. More, “Currency Recognition Using Image
Processing and Minimum Distance Classifier Technique,” Int. J. Adv.
Eng. Res. Sci., vol. 3, no. 9, pp. 1–8, 2016, doi: 10.22161/ijaers/3.9.1.
[9] B. P. Joshi and S. Kumar, A computational method of forecasting
based on intuitionistic fuzzy sets and fuzzy time series, vol. 131 AISC,
no. VOL. 2. 2012. doi: 10.1007/978-81-322-0491-6_91.
Fig. 9. Experiment Result. [10] B. P. Joshi, M. Pandey, and S. Kumar, Use of intuitionistic fuzzy time
series in forecasting enrollments to an academic institution, vol. 436.
Classification: For each category, a small number of 2016. doi: 10.1007/978-981-10-0448-3_70.
[11] S. Sharma, A. Mittal, B. Prasad Joshi, and P. Pandey, “Predictive
characteristics is identified (statistical and edge). Our method Model for Dissolved Oxygen in water body (Nainital Lake) of
computes these chosen characteristics for each category for a Nainital district by Multivariate Regression Analysis,” in 2023 3rd
given fake cash picture, then feeds them into SVM classifiers International Conference on Advance Computing and Innovative
to tell the fake from the real. SVM is a non-probabilistic binary Technologies in Engineering (ICACITE), May 2023, pp. 376–379.
linear classifier since it trains a model that categorises new doi: 10.1109/ICACITE57410.2023.10183174.
[12] B. P. Joshi, “Interval-valued intuitionistic fuzzy sets based method for

Authorized licensed use limited to: Jeju Nat'l University. Downloaded on March 01,2025 at 09:33:20 UTC from IEEE Xplore. Restrictions apply.
 multiple criteria decision-making,” Int. J. Fuzzy Syst. Appl., vol. 5, no. Three-stage Weighted 4-out-of-n:F System Subject to Possibility of
4, pp. 192–210, 2016, doi: 10.4018/IJFSA.2016100109. Degradation after Repair with Inspection,” Syst. Eng., pp. 178–206,
[13] B. Joshi, N. Joshi, and A. Gegov, “TOPSIS based Renewable-Energy- 2019, doi: 10.1201/9781351056465-9.
Source-Selection using Moderator Intuitionistic Fuzzy Set,” Int. J. [19] B. Nailwal, B. P. Joshi, and S. B. Singh, “Performance measures of a
Math. Eng. Manag. Sci., vol. 8, no. 5, pp. 979–990, Oct. 2023, doi: complex system with possible online repair,” Math. Model. Syst.
10.33889/IJMEMS.2023.8.5.056. Resil., pp. 143–170, 2019, doi: 10.1201/9781003338819-6.
[14] A. Singh, N. Kumar, B. P. Joshi, and B. K. Singh, “Load frequency [20] P. Nagaraj, V. Muneeswaran, K. Muthamil Sudar, S. Hammed, D. L.
control with time delay in restructured environment,” J. Intell. Fuzzy Lokesh, and V. Samara Simha Reddy, “An Exemplary Template
Syst., vol. 35, no. 5, pp. 4945–4951, Jul. 2018, doi: 10.3233/JIFS- Matching Techniques for Counterfeit Currency Detection,” 2022, pp.
169778. 370–378. doi: 10.1007/978-3-030-84760-9_32.
[15] B. K. Singh, N. Kumar, A. Singh, and B. P. Joshi, “BBBC based [21] S.-Y. Huang, A. Mukundan, Y.-M. Tsao, Y. Kim, F.-C. Lin, and H.-
frequency controller for hybrid power system,” J. Intell. Fuzzy Syst., C. Wang, “Recent Advances in Counterfeit Art, Document, Photo,
vol. 35, no. 5, pp. 5063–5070, Aug. 2018, doi: 10.3233/JIFS-169790. Hologram, and Currency Detection Using Hyperspectral Imaging,”
[16] A. Singh, N. Kumar, B. P. Joshi, and K. S. Vaisla, “AGC using Sensors, vol. 22, no. 19, p. 7308, Sep. 2022, doi: 10.3390/s22197308.
adaptive optimal control approach in restructured power system,” J. [22] A. Mukundan, Y.-M. Tsao, W.-M. Cheng, F.-C. Lin, and H.-C. Wang,
Intell. Fuzzy Syst., vol. 35, no. 5, pp. 4953–4962, Jul. 2018, doi: “Automatic Counterfeit Currency Detection Using a Novel Snapshot
10.3233/JIFS-169779. Hyperspectral Imaging Algorithm,” Sensors, vol. 23, no. 4, p. 2026,
[17] N. Kumar, A. Singh, B. P. Joshi, and B. K. Singh, “Intelligent Feb. 2023, doi: 10.3390/s23042026.
Adaptive Critic Scheme Implementation and Investigation Using [23] P. Dhiman, A. Kaur, C. Iwendi, and S. K. Mohan, “A Scientometric
Policy Iteration Technique: A Case Study of Multi-area Automatic Analysis of Deep Learning Approaches for Detecting Fake News,”
Generation Control (AGC) Problem,” in Lecture Notes in Electrical Electronics, vol. 12, no. 4, p. 948, Feb. 2023, doi:
Engineering, vol. 802, 2022, pp. 159–183. doi: 10.1007/978-981-16- 10.3390/electronics12040948.
6081-8_9.
[18] B. Nailwal, B. Prasad Joshi, and S. B. Singh, “Reliability Analysis of

Authorized licensed use limited to: Jeju Nat'l University. Downloaded on March 01,2025 at 09:33:20 UTC from IEEE Xplore. Restrictions apply.