Offline signature recognition and Verification system

Abstract

In the present scenario, identification and authentication of persons are an important area of research in various fields like banking transactions, documentation, etc. Biometrics play a vital role in person identification and verification compared to traditional methods.It is classified based on the physiological and behavioral characteristics of persons. Compared to other biometrics, a signature is an important and secure behavioral biometric that is widely accepted by the society for personal authentication in various applications. The proposed work is based on offline signature recognition and verification system due to its prevailing challenges compared to the online signature verification system. The proposed offline signature recognition and verification system is designed through data acquisition, pre-processing, feature extraction, feature selection, recognition and verification phases. A database of 196 signature images is created by acquiring some possible genuine signature and forged signature images from SVC20EU, ICDAR 2009 datasets and adding self- created genuine signature images. Feature extraction is an important stage in recognition and verification phases as it decides the accuracy in detecting signature forgeries. The feature selection method provides a faster verification model and increases the accuracy rate by selecting only the relevant features for classification. In this work, the hybrid features set (global, local and texture features using Gray Level Co-Occurrence Matrix (GLCM)), and texture features set (Gray Level Difference Method (GLDM) and Haar wavelets) are extracted from the created database of signatures. newline