Examinations of Offline Signature Forgery Detection from Classical to Deep Learning Techniques

Offline signature verification is an active and open research issues among the ITC applications. In the last decade the issues related to this has been evolved by traditional methods like hidden markov model, bayesian network and recently it involve deep learning approaches. However training a deep learning is very expensive in term of computation. This paper aims to develop an accurate signature forgery detection system by developing siemese network in static offline signatures, which are widely used in identity verification systems, particularly in the banking sector. However, they are vulnerable to forgeries, which can have significant consequences for legal and financial systems. In this research paper, we propose a computing similarity between two images for forgery detection. Because, traditional knowledge-based and token-based approaches to authentication are vulnerable as a person may lose the material or the information required to authenticate them. This paper will implement an optimal verification model using deep learning techniques, which will predict whether a signature is forged or not, and will also address the two individual identification of the signature owner. We will classify offline signature verification. Our research findings will contributes an accurate signature forgery detection model with enhanced security and identity verification.