Title: Streamlining check processing: advancing Arabic handwriting verification with a CNN-based system

Authors: Hamza Benyezza; Reda Kara; Mounir Bouhedda; Mosaab Benhadjer; Patrice Wira; Samia Rebouh

Addresses: Laboratory of Advanced Electronic Systems (LSEA), University of Medea, Nouveau Pôle Urbain, Medea, 26000, Algeria ' Laboratory of Advanced Electronic Systems (LSEA), University of Medea, Nouveau Pôle Urbain, Medea, 26000, Algeria ' Laboratory of Advanced Electronic Systems (LSEA), University of Medea, Nouveau Pôle Urbain, Medea, 26000, Algeria ' Laboratory of Advanced Electronic Systems (LSEA), University of Medea, Nouveau Pôle Urbain, Medea, 26000, Algeria ' IRIMAS Laboratory, University of Haute Alsace, 61 Rue Albert Camus, Mulhouse, 68200, France ' Laboratory of Experimental Biology and Pharmacology (LBPE), University of Medea, Nouveau Pôle Urbain, Medea, 26000, Algeria

Abstract: Arabic handwriting analysis and verification pose challenges due to their unique characteristics. Deep learning techniques have gained prominence in computer vision for their ability to learn from data. This study proposes a high-speed and precise solution using a convolutional neural network (CNN) to automate the verification process of Algerian postal checks written in Arabic handwriting. The solution consists of hardware and software components. The software includes four CNN models to identify the check's ID number (CID), user's signature (US), handwriting courtesy (HCA), and legal amount (HLA). The hardware setup involves a camera connected to a Raspberry PI 3. Test results demonstrate the proposed approach's effectiveness, achieving accuracy of 100% for CID, 98.61% for US, 99.28% for HCA, and 96.35% for HLA. This comprehensive system offers a promising solution for efficient verification of Arabic handwritten postal checks.

Keywords: deep learning; CNN; convolutional neural network; computer vision; Arabic handwriting; check verification.

DOI: 10.1504/IJDATS.2024.142490

International Journal of Data Analysis Techniques and Strategies, 2024 Vol.16 No.4, pp.462 - 486

Received: 21 Jun 2023
Accepted: 07 May 2024
Published online: 04 Nov 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article