Handwritten Signature Verification System using Deep Learning

Abstract

Every person has a unique signature, which is primarily used for identity documents and the authentication of significant documents or legal documentation. Static and dynamic signature verification are the two types of signature verification. The basic parameters that must be addressed when creating a signature verification system are user acceptance, the level of security necessary, accuracy, cost, and implementation. The objective of Signature verification used to determine whether a signature is real or a fake. This has proven to be a difficult endeavor, particularly in the offline scenario, which employs images of scanned signatures and does not have access to dynamic information about the signing process. The offline handwritten signature is one of the most essential biometrics used in banking systems, administrative, and financial applications. The study is to evaluate the performance of a few well-known deep convolutional neural networks as feature extractors in Handwritten Signature Verification (HSV) using transfer learning with activation function, as well as to review the available convolutional neural network signature verification methods. This is achieved by three pretrained models, VGG16, VGG19, and ResNet50, which are most often used generic models in computer vision tasks. Obtained experimental results, by comparing three models with SigComp2009 datasets and adding different parameters to each model. And conclude that VGG19 is best suitable for dataset, which has 97.83 percent accuracy than another model. These findings will help academics construct more effective Deep Learning-based signature verification methods.