Fake currency detection

inbotics Machine Learning May 7, 2023 | 0

ABSTRACT: –

Machine learning techniques help in creating applications that support fake currency detection through automated systems and algorithms. Machine learning will use pattern recognition and image processing to analyze real-world characteristics. The goal of this work is to develop a paradigm that can be supervised using related set theory and can also be useful in detecting fake datasets with very few categorization errors, thus another name referred to as a categorization model grouped as data consisting of attributes and labels denote banknotes marked as fake or genuine, and they also identify the decision boundaries that separate the samples of the two classes.

We detect the fake note from the representation of banknote characters, and some i.e the real banknote character, indicates 1 as the original and 0 as a fake note. We used the 1100 images as the dataset and among them 600 are fake notes and others are original ones. This work focuses on observing the images taken as input, which are anonymously based on the features extracted after the wavelet transformation, and also on the machine learning based problem. As a result, the processes continue by transforming the images, and we check their feasibility by testing the dataset with respect to set theory, which will be visible but not mentioned in the model. Because some properties have continuous characteristics, normalization is used to generate a set of data ranging from 0 to 1. When normalization is used, we cannot ignore or falsify any outlier features. Supervised learning helps in treating features equally and consistently differently; it is about creating a comparison model that acts as a master classifier and also helps in note detection. Several metrics are introduced focusing on feasibility and overall classification.

SYSTEM:-

data collection: The first step would be to collect a large dataset of images of both real and fake currency notes. This dataset would be used to train the machine learning model.
image preprocessing: The images would then be preprocessed to enhance their features and make them more suitable for training the model. This would involve operations like image resizing, normalization, and noise reduction .
feature extraction: The next step would be to extract meaningful features from the preprocessed images. This could involve techniques like edge detection, texture analysis, and feature point extraction.
model training:The extracted features would then be used to train a machine learning model. The model would be trained using a supervised learning approach where it would be shown examples of both real and fake currency notes and would learn to distinguish between them based on their features. model testing Once the model has been trained, it will be tested using a separate dataset of real and fake currency notes. The accuracy of the model would be evaluated based on its ability to correctly classify the notes as real or fake.
Deployments: Finally, the trained model would be deployed in a system that could be used to detect fake currency notes in real-world scenarios. This could be a mobile app that allows users to scan currency notes using their smartphone camera or a dedicated device that is used in banks and other financial institutions.

Overall, the system would provide an effective way to detect fake currency notes using machine learning, which would help to reduce financial fraud and protect the economy.

PROPOSED SYSTEM:-

In the proposed system, we will fetch the data from the Kaggle website, which contains 195 files with different Indian currencies. After fetching the dataset, we will apply some image processing techniques which are used to gather important information regarding images. In the proposed system, we work on the image of currency notes uploaded by the user. The workings of our proposed system are as follows: Firstly, we get the image of the currency note uploaded by the user. Then the RGB (colorful) image is converted into the grayscale (black and white) image. Then that grayscale image is passed through the process called edge detection.

Edge detection is a process that involves the identification of points in a digital image with discontinuities in the image brightness. This image is further processed, and edges of gray-scale images are detected. Then the image is divided into multiple parts by cropping it. Then the currency note features are cropped and segmented, and these features are stored. The intensity of each extracted feature is calculated, and if the intensity is greater than the average value, the currency note is said to be real, otherwise it is said to be fake. We mainly focused on the Security Thread, Serial Number, Latent Image, Watermark, and Identification Mark to detect whether a note is fake or not. The real currency note’s extracted features display at least 70 percent intensity; it is seen that the 500-2000 note displays intensity less than 75 percent for some features, hence it is considered a fake note.

MODULES:-

Data collection module: this module would be responsible for collecting a large dataset of images of both real and fake currency notes.
Image preprocessing module: this module would be responsible for preprocessing the collected images to enhance their features and make them more suitable for training the machine learning model. It would involve operations like image resizing, normalization, and noise reduction feature extraction modules: This module would be responsible for extracting meaningful features from the preprocessed images. It could involve techniques like edge detection, texture analysis, and feature point extraction.
model training modules: This module would be responsible for training the machine learning model using the extracted features. It would use a supervised learning approach where it would be shown examples of both real and fake currency notes and would learn to distinguish between them based on their features.
model testing module: this module would be responsible for testing the trained machine learning model using a separate dataset of real and fake currency notes. The accuracy of the model would be evaluated based on its ability to correctly classify the notes as real or fake.
deployment modules: This module would be responsible for deploying the trained machine learning model in a system that can be used to detect fake currency notes in real-world scenarios. It could be a mobile app that allows users to scan currency notes using their smartphone camera or a dedicated device that is used in banks and other financial institutions.

Overall, these modules would work together to create a robust and accurate system for detecting fake currency notes using machine learning.

APPLICATION:-

user interface : the app would have a simple and intuitive user interface that guides the user through the scanning process.
image capture: when the user is ready to scan a currency note the app would use the smartphone camera to capture an image of the note.
image preprocessing: the captured image would be preprocessed using techniques like image resizing normalization and noise reduction to enhance its features and make it more suitable for analysis.
feature extraction: the preprocessed image would be analyzed using feature extraction techniques like edge detection texture analysis and feature point extraction to extract meaningful features that can be used to distinguish between real and fake currency notes .
model prediction: the extracted features would be fed into a machine learning model that has been trained on a large dataset of real and fake currency notes the model would predict whether the scanned note is real or fake based on its learned features .
results display: the app would display the results of the analysis to the user indicating whether the scanned note is real or fake

Overall, this application would provide a simple and convenient way for users to quickly and accurately detect fake currency notes using their smartphone camera and machine learning technology.

HARDWARE AND SOFTWARE REQUIREMENTS:-

HARDWARE:-

Processor: Intel Core i3 or more.
RAM: 4GB or more.
Hard disk: 250 GB or more.

SOFTWARE:-

Operating System : Windows 10, 7, 8.
python
anaconda
Spyder, Jupyter notebook.
mysql

Fake currency detection