Volume 21, Issue 1 (3-2024)                   jor 2024, 21(1): 1-16 | Back to browse issues page

XML Persian Abstract Print

Download citation:
BibTeX | RIS | EndNote | Medlars | ProCite | Reference Manager | RefWorks
Send citation to:
Afzali P, Rezapour A, Rezaee Jordehi A. An Optimized Deep Learning Model for Handwriting Verification. jor 2024; 21 (1) :1-16
URL: http://jamlu.liau.ac.ir/article-1-2189-en.html
An Optimized Deep Learning Model for Handwriting Verification
P. Afzali , A. Rezapour * , A. Rezaee Jordehi
Department of Computer Engineering, Astaneh Ashrafieh Branch, Islamic Azad University, Astaneh Ashrafieh, Iran , Abdoreza.Rezapour@iau.ac.ir
Abstract:   (1184 Views)
Handwriting is known as a common way of communication among human societies and a behavioral biometric for authentication. Writer verification of a handwritten text has significant applications in the field of security systems, forensics and historical and literary documents. In this paper, for the first time, an optimized deep architecture utilizing Siamese network, transfer learning and gray wolf optimization algorithm is presented in order to offline writer verification from Persian handwriting. Due to the lack of access to an appropriate dataset in Persian language, a dataset including Persian words, sentences and digits has been collected. In addition, two well-known datasets of IAM and KHATT were also used to analyze the proposed method. The results of experiments and evaluations on these datasets emphasize the ability of the proposed approach to writer verification from handwriting in different languages and various writing styles.
Keywords: Writer Verification, Handwriting, Siamese Network, Transfer Learning, Gray Wolf Optimizer Algorithm
Full-Text [PDF 792 kb]   (372 Downloads)    
Type of Study: Research | Subject: Special
Received: 2023/03/23 | Accepted: 2023/08/19 | Published: 2024/03/20
Add your comments about this article : Your username or Email:
CAPTCHA
Send email to the article author

Rights and permissions
Creative Commons License This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.