Scribd
Upload
17 views

Synopsis1 1

Synopsis for Data visualisation

Uploaded by

rakshithgowdams368
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
17 views

Synopsis1 1

Synopsis for Data visualisation

Uploaded by

rakshithgowdams368
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 4

VISVESVARAYA TECHNOLOGICAL UNIVERSITY

BELAGAVI

Synopsis of

Mini Project

REAL OR FAKE CURRENCY DETECTION

Submitted by

Milan V Gowda 4BB22AI013


Darshan M C 4BB22AI003
Ullas B R 4BB22AI029

Under the Guidance of


Dr. Yashodha M, B.E., M.Tech., Ph.D.
Professor & HOD
Department of Information Science and Engineering

DEPARTMENT OF ARTIFICIAL INTELLEGENCE AND MACHINE LEARNING


Bahubali College of Engineering
Shravanabelagola – 573 135
2024-25
REAL OR FAKE CURRENCY DETECTION

AIM:

The aim of this project is to develop a Real or Fake Currency Detection system that
can accurately identify the denomination of a currency note from a set of predefined
currency images using feature matching techniques.

SCOPE:

The scope of the “Real or Fake Currency Detection” is to efficiently analyze


currency images and enhance feature detection, preprocessing steps are essential. This can
involve resizing, denoising, or adjusting contrast to highlight the unique features of the
currency notes. Once preprocessed, ORB (Oriented FAST and Rotated BRIEF) can be
employed to detect and compute keypoints and descriptors from the currency images. ORB
is a feature extraction technique well-suited for this task as it identifies critical points in the
image and generates compact, rotationally invariant descriptors. These descriptors are then
used as unique identifiers for the currency, aiding in the accurate recognition and matching
of currency features. After generating descriptors, brute force matching is utilized to find
matches between the test (unknown) and training (known) currency images. This method
compares each descriptor in the test image to those in the training set, measuring similarity
to find potential matches. Lowe's ratio test is applied next to refine these matches by
ensuring only the most distinct, high-quality matches are retained, reducing the risk of
incorrect identifications. Based on the evaluation, the currency note is either displayed as
matched, indicating genuine currency, or flagged if no reliable match exists, which could
indicate counterfeit currency. This process provides a robust approach to determining
authenticity in currency detection systems.

OBJECTIVES:
 To preprocess currency note images using techniques such as resizing, grayscale
conversion, bilateral filtering, and morphological processing.
 To extract and match key features from currency note images using the ORB
(Oriented FAST and Rotated BRIEF) detector.
 To implement a matching algorithm using the BF (Brute Force) matcher and
Lowe's ratio test for identifying good matches.
METHODOLOGY:

1. Image Acquisition and Preprocessing

 Use high-resolution cameras or scanners to capture images of the currency. The


resolution and lighting should be sufficient to reveal details like color, texture,
and small markings.

 This step includes noise reduction, grayscale conversion, and contrast


adjustment to enhance the visibility of distinguishing features. Cropping or
resizing may be used to ensure consistent dimensions.

2. Feature Extraction

 Different currencies have specific colors. Image processing algorithms analyze


RGB or HSV color models to differentiate currencies based on unique color
schemes.

 Many currencies have specific textures detectable through methods like Gabor
filters, Local Binary Patterns (LBP), or wavelet transforms. These methods
assess textures to capture patterns that distinguish between currencies.

 Edge detection algorithms, such as the Canny or Sobel edge detectors, help
identify shapes and figures unique to each currency.

 Optical Character Recognition (OCR) is applied to identify specific numbers,


symbols, or text (e.g., denomination numbers or country names) on the
currency.

3. Pattern Recognition Using Machine Learning Models

 For simpler systems, template matching compares the captured image to stored
templates of each currency type and denomination. Correlation metrics measure
similarity.

 For higher accuracy, machine learning models like Convolutional Neural


Networks (CNNs) are trained on labeled images of different currencies. The
CNN extracts and learns features that distinguish between currencies.

4. Validation and Confidence Scoring

 The system assigns a confidence score to its recognition, which is critical for
reducing false positives.

 Based on confidence scores, the system may reject uncertain classifications or


request additional input (e.g., retake the image).
5. Output and Decision-Making

 Based on the highest confidence score, the currency and its denomination are
identified.

6. Testing and Iteration

 The model is tested with a variety of currencies, light conditions, and angles.

 Feedback and retraining with more diverse datasets improve accuracy over
time.

HARDWARE AND SOFTWARE REQUIREMENTS:


 Processor: A quad-core Intel Core i7

 RAM: At least 8GB

 Storage: A minimum of 500GB HDD/ SSD

 Graphic card: Nvidia 9x or 10x series

 Tkinter

 Numpy

 Matplotlib

 Jupyter Notebook

RESULT EXPECTED:
The program should correctly identify the denomination of the test currency note and
display the matched image. If no good match is found, it should indicate "FAKE
CURRENCY".

Signature of the students:

Milan V Gowda 4BB22AI013

Darshan M C 4BB22AI003

Ullas B R 4BB22AI029

Dr. Yashodha M Smt. Veena Nyamagoud Mrs. Shylaja L N


Professor & Guide Asst. Professor & Coordinator Asso. Professor & HOD
Dept. of IS & E Dept. of IS & E Dept. of AI & ML

You might also like