A Thresholded Gabor-CNN Based Writer Identification System for Indic Scripts

M F Mridha,Muhammad Mohsin Kabir,Jungpil Shin,Md Abdul Hamid,Muhammad Mostafa Monowar,Abu Quwsar Ohi

doi:10.1109/access.2021.3114799

Abstract

Writer identification is the procedure of identifying individuals from handwriting. Writer identification is a common interest in biometric authentication and verification systems, and numerous studies are available for English, Chinese, Arabic, and Persian specific handwriting. This paper introduces a supervised offline Indic script writer identification system that can identify individuals using less than a single page of handwriting. A lightweight Convolutional Neural Network (CNN) architecture fused with non-trainable Gabor filters is used as an identification model that can recognize writers from scarce data. For the experiment, we used BanglaWriting dataset, which is openly available for Bengali writing and writer recognition. Further, we added Devanagari and Telugu datasets for evaluation. The overall evaluation shows that the proposed thresholded Gabor-based CNN architecture performs superior to numerous deep CNN architectures for Indic writer recognition.

Highlights

H ANDWRITING biometrics is a division of biological phenomena that allows a person to be identified and authenticated based on a set of recognizable and verifiable features that are unique and specific among individuals [1]
We introduce a Deep Convolutional Neural Network (DCNN) based writer recognition system, for Indic script writer recognition systems
The writer identification system uses a traditional CNN architecture that is fused with a modified Gabor filter

Summary

Introduction

H ANDWRITING biometrics is a division of biological phenomena that allows a person to be identified and authenticated based on a set of recognizable and verifiable features that are unique and specific among individuals [1]. The usage of an online writer identification system is limited to electronic writing devices, offline writer identification can be applied to both electronic and paper-based handwriting. Writer identification systems are segmented into two categories based on the criterion of recognition procedure: text-dependent and text-independent recognition [3]. Textdependent recognition systems can only recognize writers from a specific written word phrase which the system has previously seen as an example from the writer (or stored in the database). Signature verification systems are a variant of text-dependent writer recognition. In such a case, writers provide specific signatures which they have to write again for validation purposes. Text-independent systems can recognize any writing from the writers, which the system may not have previously observed by the writer. By writer recognition, we term the text-independent writer recognition systems

Methods

Results

Conclusion