SIGNATURE VERIFICATION USING CNN

"Signature verification can be considered a special case of pattern recognition. Like in any pattern recognition problem, in signature verification distinctive features can be extracted from a set of original signatures. However, Approaches to signature verification. There are two types of signatures, offline (static) & online (dynamic).

Online signatures have higher distinctive features but offline signatures have fewer distinctive features. With the development of machine learning, new algorithms present promising solutions that can be used for signature verification. For these reasons, signature verification is one of the most important problems remains to be solved in machine learning methods nowadays."

Arun Kumar Sahu

Visit GitHub

arunkumarsahu634@ gmail.com

Visit LinkedIn Profile

TECHNICAL

"Some additional text related to the Project."

Components

Language
PYTHON
Machine Learning

library
Tkinter
CV2
Numpy
CSV
Pandas
Matplotlib

Algorithm
Convolutional Neural Network (CNN)

Platform Environment
Visual Studio Code

Purpose

Traditionally signature was manually compared with copies of genuine signatures for verification. This simple method may not be sufficient as the technology is becoming more and more advance and with advancing techniques of forgeries and falsification of signature. So, in order to tackle such problem new efficient tool is needed and this project proposes such signature verification tool which can assist human in correct decision making in authentication of handwritten signature.

For such authentication of signature this project presents an applications of which facilitates the feature of human signature verification using the convolution neural network approach. This software is able to train the network with new dateset of signature and validate the authenticity of new signature of trained class.

Problem Solving

In this study we will take the dataset of the different signatures. We will take signature as an input in the form of image. After taking signature as an input next step is feature extraction. Signature is separated according to the features. After extracting the features, matching process is done. According to the matching result is implemented. And final output is recognition of signature.

Main motivation of the system is to provide security to the signatures which are widely used in legal documents, banking and commercial transactions. Normally it is difficult to identify the signature of the particular user at the time of verification of the documents. At that time, there is a need to identify the fake signatures from the documents. So there must be some system or application which would help banking system and some other systems like commercial transaction to detect the fake signature.