MACHINE LEARNING BASED FAKE CURRENCY DETECTION SYSTEM

Abstract

ABSTRACT A MACHINE LEARNING BASED FAKE CURRENCY DETECTION SYSTEM The present invention relates to a machine learning based fake currency detection system [100] comprising an input module [102] with a camera and multiple image sensors, designed to capture detailed images of banknotes, for covering both visible and non-visible features. Furter, a pre-processing unit [104] enhances these images, allowing an artificial intelligence (AI) or machine learning (ML) module [106] to identify key parameters and extract textural details. The AI/ML module [106] further identifies patterns unique to authentic currency. The AI/ML module [106] trains at least one model trained with data augmentation techniques, such as elastic distortions, geometric transformations, and lighting changes, for an accurate recognition of real and counterfeit banknotes. Further, the at least one trained model is then deployed to distinguish genuine currency from counterfeits in real-time applications. Refer to Figure 1.

Specification

Description:"MACHINE LEARNING BASED FAKE CURRENCY DETECTION SYSTEM"
FIELD OF THE INVENTION
[0001] The present invention relates to the field of fake currency detection system, which specifically focuses on techniques for verifying the authenticity of currency notes. The present invention uses image processing, feature extraction, and deep learning models to analyse security features and textural patterns in currency, aiming to identify counterfeit notes with high accuracy. The present invention addresses challenges in real-time, efficient, and reliable currency verification, applicable across banking, retail, and financial sectors.
BACKGROUND OF THE INVENTION
[0002] In present, counterfeit currency poses a significant threat to economies worldwide. The circulation of phony currency harms the value of real money, leading to market fluctuations, disruptions in commerce, and inflation. These issues directly impact consumers, businesses, and financial institutions. Also, the public trust in financial institutions is reduced overtime, when the counterfeit currency spreads, as individuals become uncertain about the authenticity of the money that are used by them on a daily purpose.
[0003] Despite the existence of security features on currency, counterfeiters continually find ways to replicate them, creating an ongoing need for advanced, reliable detection systems. Therefore, the present disclosure addresses these challenges and is focused on providing a solution that is able to identify counterfeit currency, in order to assist in financial stability and public confidence.

SUMMARY OF THE INVENTION
[0004] In view of the foregoing disadvantages inherent in the prior art, the general purpose of the present disclosure is to provide a, to include all advantages of the prior art, and to overcome the drawbacks inherent in the prior art.
[0005] Some of the objects of the present disclosure, which at least one embodiment herein satisfies, are as follows:
[0006] An object of the present disclosure is to provide a system for accurately detecting counterfeit currency notes with high precision in diverse environmental conditions. An object of the present disclosure is to provide a solution capable of identifying subtle differences between authentic and counterfeit currency, ensuring reliable detection.
[0007] Another object of the present disclosure is to provide a system that operates efficiently in real-time, optimizing resource use for fast and effective counterfeit detection.
[0008] Another object of the present disclosure is to provide a system that is adaptable to multiple currencies, making it versatile for global application.
[0009] Another object of the present disclosure is to offer a counterfeit detection solution with low computational overhead, minimizing hardware requirements and enabling cost-effective implementation.
[0010] Yet another object of the present disclosure is to provide a system suitable for integration into various sectors, including banking, retail, and government, for widespread use.
[0011] Other objects and advantages of the present disclosure will be more apparent from the following description, which is not intended to limit the scope of the present disclosure.
[0012] The modular mobility device for patients of the present disclosure facilitates an input module with a camera and multiple image sensors, designed to capture detailed images of banknotes, for covering both visible and non-visible features. Furter, a pre-processing unit enhances these images, allowing an artificial intelligence (AI) or machine learning (ML) module to identify key parameters and extract textural details. The AI/ML module further identifies patterns unique to authentic currency. The AI/ML module trains at least one model trained with data augmentation techniques, such as elastic distortions, geometric transformations, and lighting changes, for an accurate recognition of real and counterfeit banknotes. Further, the at least one trained model is then deployed to distinguish genuine currency from counterfeits in real-time applications.

BRIEF DESCRIPTION OF DRAWING
[0013] The foregoing summary, as well as the following detailed description of various embodiments, is better understood when read in conjunction with the drawings provided herein. For the purposes of illustration, there are shown in the drawings exemplary embodiments; however, the presently disclosed subject matter is not limited to the specific methods and instrumentalities disclosed.
[0014] Figure 1 illustrates an exemplary block diagram of a machine learning based fake currency detection system, is shown, in accordance with exemplary implementations of the present disclosure; and
[0015] Figure 2, illustrates an exemplary method flow diagram for operating the machine learning based fake currency detection system, in accordance with exemplary implementations of the present disclosure.
[0016] Like reference numerals refer to like parts throughout the description of several views of the drawing.
DETAILED DESCRIPTION OF THE INVENTION
[0017] Embodiments are provided so as to thoroughly and fully convey the scope of the present disclosure to the person skilled in the art. Numerous details are set forth, relating to specific components, and methods, to provide a complete understanding of embodiments of the present disclosure. It will be apparent to the person skilled in the art that the details provided in the embodiments should not be construed to limit the scope of the present disclosure. In some embodiments, well-known processes, well- known apparatus structures, and well-known techniques are not described in detail.
[0018] The terminology used, in the present disclosure, is only for the purpose of explaining a particular embodiment and such terminology shall not be considered to limit the scope of the present disclosure. As used in the present disclosure, the forms "a," "an," and "the" may be intended to include the plural forms as well, unless the context clearly suggests otherwise. The terms "comprises," "comprising," "including," and "having," are open-ended transitional phrases and therefore specify the presence of stated features, integers, steps, operations, elements, modules, units and/or components, but do not forbid the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof. The particular order of steps disclosed in the method and process of the present disclosure is not to be construed as necessarily requiring their performance as described or illustrated. It is also to be understood that additional or alternative steps may be employed.
[0019] The following detailed description should be read with reference to the drawings, in which similar elements in different drawings are identified with the same reference numbers. The drawings, which are not necessarily to scale, depict illustrative embodiments and are not intended to limit the scope of the disclosure.
[0020] It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention, as claimed. In this application, the use of the singular includes the plural, the word "a" or "an" means "at least one", and the use of "or" means "and/or", unless specifically stated otherwise. Furthermore, the use of the term "including", as well as other forms, such as "includes" and "included", is not limiting. Also, terms such as "element" or "component" encompass both elements and components comprising one unit and elements or components that comprise more than one unit unless specifically stated otherwise.
[0021] Furthermore, the term "module", as used herein, refers to logic embodied in hardware or firmware, or to a collection of software instructions, written in a programming language, such as, Java, C, C++, python, or assembly. One or more software instructions in the modules can be embedded in firmware, such as in an EPROM. The modules described herein can be implemented as either software and/or hardware modules and can be stored in any type of non-transitory computer-readable medium or other storage device. Some non-limiting examples of non-transitory computer-readable media include CDs, DVDs, BLU-RAY, flash memory, and hard disk drives.
[0022] Referring to FIG. 1, an exemplary block diagram of a machine learning based fake currency detection system [100], is shown, in accordance with exemplary implementations of the present disclosure. The machine learning based fake currency detection system [100] comprises at least one input module [102], at least one pre-processing unit [104], and at least one artificial intelligence/ machine learning (AI/ML) module [106]. Also, all of the components of the machine learning based fake currency detection system [100] are assumed to be connected to each other unless otherwise indicated below. As shown in the figures all components shown within the fake currency detection system [100] should also be assumed to be connected to each other. Also, in FIG. 1, only a few units are shown, however, the fake currency detection system [100] may comprise multiple such units or the fake currency detection system [100] may comprise any such numbers of said components, as required to implement the features of the present disclosure.
[0023] The fake currency detection system [100] comprises the input module [102] having a camera and a plurality of image sensors. Herein, the image sensors include a combination of optical, infrared, and UV sensors. Further, the input module [102] is configured to capture one or more images of one or more bank notes.
[0024] The camera herein may capture high-resolution images of banknotes which may include a full, clear image of each banknote, along with the layout, texture, and any visual details present on each banknote. Further, the optical sensor captures visible light features of the banknote such as the colors, contrasts, and patterns of the banknote. Also, the infrared sensor captures non-visible light features of the banknote by detecting infrared light which is useful in identifying material present on the banknote that are not visible under standard lighting. Further, the UV sensor detects features visible under ultraviolet light, such as examining fluorescent inks and special markings that may present on the banknote. The input module [102] with the combination of the camera and the plurality of image sensors may capture a multispectral profile of the banknote.
[0025] The fake currency detection system [100] further comprises the pre-processing unit [104] configured to perform an image acquisition to identify one or more parameters associated with the one or more bank notes. The pre-processing unit [104] retrieves raw image data of each banknote in a plurality of lighting conditions such as visible light, UV, IR, in order to capture a complete and clear representation of the banknote. Herein, the one or more parameters include but not limited to watermarks, security threads, and microprints during the picture markup phase.
[0026] In an implementation, the watermarks may have distinctive opacity and contrast levels under different lighting conditions, therefore the pre-processing unit [104] may identify watermarks by analyzing transparency and light diffusion patterns within the retrieved image. In another implementation, the threads often have IR- or UV-responsive properties, therefore the re-processing unit identifies security threads by analyzing the image for specific linear or patterned elements within the banknote. In yet another implementation, the pre-processing unit [104] identifies the microprints through high-resolution image analysis such as extremely magnified sections of the banknote image. In yet another implementation, the pre-processing unit [104] may highlight specific key features in the banknote for further processing.
[0027] The fake currency detection system [100] further comprises the AI/ML module [106] for receiving the one or more images from the input module [102]. Post highlight specific key features in the banknote by the pre-processing unit [104], the images are passed to the AI/ML module [106]. Post receiving the images, the AI/ML module [106] extracts feature from the received image. Herein, the feature includes one or more textural features of the one or more bank notes. Also, the textural features extracted from the image include one or more of roughness, periodicity, directionality, and coarseness The AI/ML module [106] firstly converts the received image from RGB to grayscale, then the AI/ML module [106] utilizes a Texture Co-occurrence Matrix (TLCM) to quantify spatial patterns by analyzing how often pairs of pixel intensities (gray levels) occur at certain distances and angles.
[0028] Further, by analyzing pixel intensity variations using the TLCM, the AI/ML module [106] is able to identify roughness by detecting larger intensity differences, which typically signify a rougher texture. Further, the TLCM may observe repeating patterns, as recurring pixel intensity pairs indicate periodicity. In addition, the TLCM calculates pixel intensity relationships in specific directions, such as horizontal or diagonal, which assist the AI/ML module [106] to recognize patterns with specific orientations. Moreover, the TLCM assesses coarseness by observing the distribution of pixel intensity pairs at various distances.
[0029] Further, the AI/ML module [106] is configured to apply recurrent neural networks (RNNs) with a multi-focal attention mechanism to process both local and global components of the banknote, and segmentation networks to split the images into brightness levels for identifying distinct regions and objects. The RNNs are designed to process sequential data by retaining contextual information through feedback loops, which is particularly useful in analyzing the sequential spatial details across the image of a banknote. Herein, the multi-focal attention mechanism allows the RNNs to dynamically focus on different areas of the banknote with varying levels of detail for identifying distinct regions and objects. In one aspect, the multi-focal attention mechanism focuses on smaller regions of the banknote, enabling the RNN to capture minute details such as microprints and fine textures. In another aspect, the multi-focal attention mechanism focusses on broader regions, allowing the RNN to capture the overall layout, design symmetry, and structural elements of the banknote.
[0030] Further, the AI/ML module [106] is configured to utilize a multi-layered Convolutional Neural Network (CNN) with Softmax activation to classify the segmented features as real or fake based on the extracted textural features. Herein, the segmentation of the banknote divides the banknote image into distinct regions based on variations in brightness and other visual characteristics. The multi-layered CNN consists of several convolutional layers that apply filters to the segmented images to detect and analyze textural and spatial patterns within the banknote. As, each layer of the CNN extracts increasingly complex features by identifying edges, textures, shapes, and patterns within the segmented regions of the banknote, the CNN is able to distinguish elements that are characteristic of genuine currency from those typically absent or poorly replicated in counterfeits. Further, based on the extracted features the CNN uses Softmax to categorize the segmented features as real or fake, yielding a probability value for each. For instance, a high probability for "real" would indicate that the analyzed textures and patterns align closely with those of an authentic banknote.
[0031] Post classification, the AI/ML module [106] applies post-processing techniques to further refine the image quality and reduce noise. In one aspect, the post-processing technique may include a median filtering technique, which is a noise reduction technique that replaces each pixel's value with the median value of neighboring pixels. In another aspect, the post-processing technique may include a morphological analysis technique, which is used to refine the shapes and structures within the image.
[0032] The AI/ML module [106] is further configured to train at least one model based on the identified one or more parameters and from the one or more images and extracted features. Further, the at least one model is trained using data augmentation techniques such as elastic distortions, geometric transformations, and lighting modifications. In an implementation, the elastic distortions simulate stretching or warping effects, allowing the at least one model to recognize authentic banknotes even when their images are altered due to wear, tear, or handling. In another implementation, the geometric transformations may include transformations, such as scaling, shearing, and perspective changes, enable the at least one model to recognize currency notes viewed from different angles or under various conditions. In yet another implementation, the lighting modifications may imply that variation in lighting simulate different environmental conditions, helping the at least one model to distinguish real banknotes in various light levels or shadowed conditions. Further, the data augmentation techniques comprise random cropping, flipping, and rotation of the captured images to simulate different orientations and distortions of the currency notes. Post training of the at least one model, the at least one model is deployed to perform real-time classification, determining whether a banknote is authentic or counterfeit.
[0033] Referring to FIG. 2, an exemplary method flow diagram [200] for operating the machine learning based fake currency detection system, in accordance with exemplary implementations of the present disclosure is shown. In an implementation the method [200] is performed by the fake currency detection system [100].
[0034] Also, as shown in Figure 2, the method [200] initially starts at step [202].
[0035] At step [204], the method [200] comprises capturing, via an input module [102] having the camera and the plurality of image sensors, one or more images of one or more banknotes. The camera may capture high-resolution images of banknotes which may include a full, clear image of each banknote, along with the layout, texture, and any visual details present on each banknote. Further, the optical sensor captures visible light features of the banknote such as the colors, contrasts, and patterns of the banknote.
[0036] At step [206], the method [200] comprises performing, via the pre-processing unit [104], an image acquisition to identify one or more parameters associated with the one or more bank notes. The pre-processing unit [104] retrieves raw image data of each banknote in a plurality of lighting conditions such as visible light, UV, IR, in order to capture a complete and clear representation of the banknote. Herein, the one or more parameters include but not limited to watermarks, security threads, and microprints during the picture markup phase.
[0037] At step [208], the method [200] comprises receiving, via the AI/ML module [106] communicatively coupled with the input module [102], the one or more images. Post highlight specific key features in the banknote by the pre-processing unit [104], the images are passed to the AI/ML module [106].
[0038] At step [210], the method [200] comprises extracting, via the AI/ML module [106], features from the segmented image, including one or more textural features of the one or more bank notes by converting the image from RGB to grayscale and utilizing a Texture Co-occurrence Matrix (TLCM). The textural features extracted from the image include one or more of roughness, periodicity, directionality, and coarseness. Herein, the AI/ML module [106] firstly converts the received image from RGB to grayscale, then the AI/ML module [106] utilizes a Texture Co-occurrence Matrix (TLCM) to quantify spatial patterns by analyzing how often pairs of pixel intensities (gray levels) occur at certain distances and angles. Further, the AI/ML module [106] is configured to apply recurrent neural networks (RNNs) with a multi-focal attention mechanism to process both local and global components of the banknote, and segmentation networks to split the images into brightness levels for identifying distinct regions and objects.
[0039] At step [212], the method [200] comprises training, via the AI/ML module [106], at least one AI model based on the identified one or more parameters and from the one or more images and extracted features, wherein the at least one model is trained using data augmentation techniques such as elastic distortions, geometric transformations, and lighting modifications. The method [200] further comprises comparing, via the AI/ML module [106], the extracted textual features with a database of known genuine bank note textures for classification.
[0040] At step [214], the method [200] comprises deploying, via the AI/ML module [106], the trained model to determine an authentic or counterfeit bank note. Post training of the at least one model, the at least one model is deployed to perform real-time classification, determining whether a banknote is authentic or counterfeit.
[0041] The method [200] herein terminates at step [216].

[0042] While considerable emphasis has been placed herein on the specific features of the preferred embodiment, it will be appreciated that many additional features can be added and that many changes can be made in the preferred embodiment without departing from the principles of the disclosure. These and other changes in the preferred embodiment of the disclosure will be apparent to those skilled in the art from the disclosure herein, whereby it is to be distinctly understood that the foregoing descriptive matter is to be interpreted merely as illustrative of the disclosure and not as a limitation.
[0043] While the invention has been described in connection with what is presently considered to be the most practical and various embodiments, it is to be understood that the invention is not to be limited to the disclosed embodiments, but on the contrary, is intended to cover various modifications and equivalent arrangements.
[0044] The embodiments described above are intended only to illustrate and teach one or more ways of practicing or implementing the present invention, not to restrict its breadth or scope. The actual scope of the invention, which embraces all ways of practicing or implementing the teachings of the invention, is defined only by the following claims and their equivalents.
, Claims:I/We Claim:
1. A machine learning based fake currency detection system [100], comprising:
an input module [102] having a camera and a plurality of image sensors, configured to capture one or more images of one or more bank notes;
a pre-processing unit [104] configured to perform an image acquisition to identify one or more parameters associated with the one or more bank notes;
an artificial intelligence (AI)/machine learning (ML) module [106] communicatively coupled with the input module [102], wherein the AI/ML module [106] is configured to:
receive the one or more images,
extract features from the received image, including one or more textural features of the one or more bank notes by converting the image from RGB to grayscale and utilizing a Texture Co-occurrence Matrix (TLCM),
train at least one model based on the identified one or more parameters and from the one or more images and extracted features, wherein the at least one model is trained using data augmentation techniques such as elastic distortions, geometric transformations, and lighting modifications, and
deploy the trained model to determine an authentic or counterfeit bank note.
2. The machine learning based fake currency detection system [100] as claimed in claim 1, wherein the one or more parameters include but not limited to watermarks, security threads, and microprints during the picture markup phase.

3. The machine learning based fake currency detection system [100] as claimed in claim 1, wherein the AI/ML module [106] is configured to apply recurrent neural networks (RNNs) with a multi-focal attention mechanism to process both local and global components of the banknote, and segmentation networks to split the images into brightness levels for identifying distinct regions and objects.

4. The machine learning based fake currency detection system [100] as claimed in claim 3, wherein the AI/ML module [106] is configured to utilize a multi-layered Convolutional Neural Network (CNN) with Softmax activation to classify the segmented features as real or fake based on the extracted textural features.

5. The machine learning based fake currency detection system [100] as claimed in claim 1, wherein the AI/ML module [106] is configured to apply post-processing techniques such as median filtering and morphological analysis to smooth the image and reduce noise.

6. The machine learning based fake currency detection system [100] as claimed in claim 1, wherein the image sensors include a combination of optical, infrared, and UV sensors to capture both visible and non-visible features of the bank notes.

7. The machine learning based fake currency detection system [100] as claimed in claim 1, wherein the data augmentation techniques further comprise random cropping, flipping, and rotation of the captured images to simulate different orientations and distortions of the currency notes.

8. The machine learning based fake currency detection system [100] as claimed in claim 1, wherein the textural features extracted from the image include one or more of roughness, periodicity, directionality, and coarseness.

9. A method [200] for operating the machine learning based fake currency detection system [100] as claimed in claim 1, wherein the method [200] comprising:

capturing, via an input module [102] having a camera and a plurality of image sensors, one or more images of one or more bank notes;
performing, via a pre-processing unit [104], an image acquisition to identify one or more parameters associated with the one or more bank notes;
receiving, via an artificial intelligence (AI) module/ machine learning (ML) module [106] communicatively coupled with the input module [102], the one or more images;
extracting, via the AI/ML module [106], features from the segmented image, including one or more textural features of the one or more bank notes by converting the image from RGB to grayscale and utilizing a Texture Co-occurrence Matrix (TLCM),;
training, via the AI/ML module [106], at least one AI model based on the identified one or more parameters and from the one or more images and extracted features, wherein the at least one model is trained using data augmentation techniques such as elastic distortions, geometric transformations, and lighting modifications; and
deploying, via the AI/ML module [106], the trained model to determine an authentic or counterfeit bank note.
10. The method [200] for operating the machine learning based fake currency detection system [100] as claimed in claim 9, further comprising comparing, via the AI/ML module [106], the extracted textual features with a database of known genuine bank note textures for classification.

Documents