Application of Convolutional Neural Network for Forgery detection in Handwritten Signatures

Abstract

Offline signature forgery has become increasingly common, making identification and verification essential for security and resource access control. Forgery can be divided into three types: 1. Random forgery2. Simple or casual forgery, and 3. expert or Skilled (or simulated) forgery. The primary goal of signature forgery detection is to extract distinctive handwriting features or patterns from the signature and determine whether it is genuine or forged. Signature verification can be classified into: static (offline) and dynamic (online). In our proposed solution, we focus on offline signature analysis for forgery detection. This process involves

capturing the signature image and applying image pre-processing techniques to enhance its quality for further analysis Feature extraction algorithms are used to identify and extract essential characteristics from signature images. These extracted features serve as input parameters for a machine learning algorithm, which is used to differentiate between genuine and forged signatures. Once the analysis is complete, performance evaluation is conducted to assess the accuracy of the results.[1][5]