Literature Paper on Offline Signature Verification

Abstract

The effectiveness of comparison features classified by humans for handwritten semi-automatic signature recognition is examined in this paper. Despite the widespread use of automated technology, people are still crucial to many jobs involving the recording and verification of handwritten signatures. The main focus of several study lines is how humans may contribute to the improvement of these systems. Comparative attributes make an effort to take use of our capacity for extracting discriminatory information from signatures. Comparative characteristics provide deeper responses than absolute attributes (such as "is this stroke vertical?"), such as "how vertical is this stroke?" In this study, we introduce a new interface for semiautomatically classifying signatures that was inspired by forensic document examiners (FDE). A new Bio-HSL (Biometric-Handwritten Signatures Labeling) database with 4,968,600 signature attributes is created through the manual labelling process. The findings demonstrate that for semi-automatic signature recognition, comparative attributes outperform absolute attributes, with Equal Error Rates ranging from 5.5% for random comparisons to 21.2% for simulated forgeries.