Identification Identification of Signature Writing Using Convolutional Neural Network Algorithm

Abstract

Signature is the outcome of a writing process with distinct characteristics, serving as one of the proofs of the validity of an agreement conducted by two or more parties as evidence of identity verification. This study aims to design a system capable of identifying an individual based on inputted signature images into the system. The rapid development of knowledge can certainly be leveraged to facilitate and address issues in human daily life, one of which is the application of expertise in the field of pattern recognition, enabling the creation of a signature identification system. There are five main stages employed in this research, namely image acquisition, image augmentation, system architecture design, training process, and testing process. The research results demonstrate that the applied method proves to be effective in designing a signature identification system. This is substantiated by the accuracy level of the system testing reaching 98.148%.