International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 11 Issue: 03 | Mar 2024
p-ISSN: 2395-0072
www.irjet.net
DUPLICATE SIGNATURE VERIFICATION USING CNN ALGORITHM Mrs.I.Suganya,M.E.,MBA., K.Sai krishna
D. Sai ramesh
Yenamala vamsi
Assistant professor Student Department of Artificial Department of Artificial and data science intelligence and data science Muthayammal engineering college. Muthayammal engineering college Rasipuram. Rasipuram.
Student Department of Artificial intelligence and data science Muthayammal engineering college Rasipuram.
Student Department of Artificial Intelligence intelligence and data science Muthayammal engineering college Rasipuram.
---------------------------------------------------------------------***-------------------------------------------------------------------
Abstract: Many different security programs use biometric technology. These systems try to identify people based on physical or behavioral characteristics. In the first case, identification is based on the measurement of biological features such as fingerprints, faces, and irises. The second scenario includes behavioral features such as speech and handwriting. Authentication and identification are two basic applications of biometric technology. In the first scenario, a system user provides a biometric sample while claiming who he is. The function of an authentication system is to determine whether the user really is who they say they are. When a user provides a biometric sample, the goal of the identification scenario is to find that user's biometric sample among other users registered in the system. This project introduces an innovative approach for duplicate signature verification using Convolutional Neural Networks (CNN). Signature verification is an important aspect of document authentication and fraud prevention. The proposed system uses CNN, a powerful class of deep learning algorithms suitable for image-based tasks, to analyze and distinguish between real and duplicate signatures. The system starts by capturing high-resolution images of signatures that incorporate different writing styles and conditions. These images serve as input to a CNN architecture, which is trained to learn complex features and patterns that represent valid signatures. The trained model is used for real-time verification to effectively distinguish between genuine and duplicate signatures. Extensive experiments and validation demonstrate the robustness and accuracy of the CNN-based approach in detecting duplicate signatures. Keywords: convolutional neural network;
Higher accuracy. The main advantage of signature verification systems over other types of technology is that signatures are already accepted as a common method of identity verification. There are two types of handwritten signature verification: online and offline. Online methods use electronic techniques and computers to extract signature information and obtain dynamic information such as pressure, speed, and writing speed for verification purposes. Offline signature verification uses fewer electronic controls and a signature image captured with a scanner or camera. Offline signature verification systems use features extracted from scanned signature images. The function used for offline signature verification is very simple. Here only pixel images should be evaluated. However, offline systems are difficult to design because many desirable features such as stroke order, velocity, and other dynamic information are not available offline. The verification process should be based entirely on the features that can be extracted from the static signature trace. Problem Statement : Handwritten signatures are unique to each person and impossible to duplicate. This technology is easy to explain and reliable. Identifying and verifying handwritten signatures from images is a big problem. This is very difficult because the human eye does not have the visual ability to recognize all the details of handwriting. It is difficult for humans to sign because the signature changes every time Verification of handwritten signatures There are two types: online and offline. Online methods use electronic devices. Technology and computers to extract information about signatures and photographs Dynamic information such as pressure, speed and writing speed for targets Offline signature verification requires fewer electronic controls
Introduction : For legal transactions, authorization is by signature. This increases the need for signature verification. Handwritten signatures are unique to each person and impossible to duplicate. This technology is easy to explain and reliable. Identifying and verifying handwritten signatures from images is a big problem. This is very difficult because even the human eye does not have the visual ability to detect all internal details. handwritten. Since the signature changes every time, it is difficult for humans to distinguish between genuine and fake products. Deep learning using a highly digital replica of the human brain can be used to identify fake signatures.
© 2024, IRJET
|
Impact Factor value: 8.226
Use a signature image taken with a scanner or camera. Offline signature The verification system uses features extracted from the scanned signature image. From The function used for offline signature verification is very simple. only this, I need to evaluate pixel images. However, offline systems are difficult to design Provide as many desirable features as possible, such as stroke order, speed, and impact. In offline mode, no other dynamic information is available. Confirmation This process should be based entirely on features that can be extracted from the trace. Static signature image
|
ISO 9001:2008 Certified Journal
|
Page 493