EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Designing a Novel CNN–LSTM-based Model for Arabic Handwritten Character Recognition for the Visually Impaired Person

Manel Ayadi, Nesrine Masmoudi, Latifa Almuqren, Hadeel Saeed Alshahrani, Raneem Oudah Aljohani

Journal of Disability Research · 2025 · 4 (1)

DOI: 10.57197/jdr-2024-0080 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Disability Research, 2025. Verfasst von Manel Ayadi und 4 weiteren Autor:innen. Die Arbeit wurde bisher 4-mal zitiert (Crossref) und nennt 42 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The field of healthcare has undergone a radical change as a consequence of the latest advancements in deep learning. Recently, the development of visual substitutes for visually impaired people (VIP) has significantly aided research on assistive technology (AT). However, there is still little usage of ATs to understand the fundamental meaning of various written substances. This research presents a novel hybrid model of convolutional neural network (CNN) and long short-term memory (LSTM) for Arabic handwritten character recognition (AHCR) to present AT for VIP. This technique offers a practical way to improve accessibility for those who are visually impaired. The CNN’s convolutional layers are used to capture both local and global patterns by extracting hierarchical information from the intricate and varied shapes in Arabic characters. After feeding these features into the LSTM network, the model comprehends the sequential nature of Arabic writing by capturing contextual information. Combining these two architectures allows the model to take advantage of temporal as well as spatial correlations, which improves recognition accuracy for complex Arabic letters. In this work, text-to-speech technology is also used to turn the recognized text into audio. To validate the model performances, we employed the publically available Arabic Handwritten Characters Dataset, which includes a range of writing situations and styles. The proposed CNN–LSTM model outperforms conventional methods for AHCR and achieves the highest accuracy of 98.07% over the state-of-the-art approaches.

Zitieren

APA

Ayadi, M., Masmoudi, N., Almuqren, L., Alshahrani, H. S., & Aljohani, R. O. (2025). Designing a Novel CNN–LSTM-based Model for Arabic Handwritten Character Recognition for the Visually Impaired Person. Journal of Disability Research, 4(1). https://doi.org/10.57197/jdr-2024-0080

Vancouver

Ayadi M, Masmoudi N, Almuqren L, Alshahrani HS, Aljohani RO. Designing a Novel CNN–LSTM-based Model for Arabic Handwritten Character Recognition for the Visually Impaired Person. Journal of Disability Research. 2025;4(1). doi:10.57197/jdr-2024-0080

BibTeX
@article{Ayadi2025,
  author = {Ayadi, Manel and Masmoudi, Nesrine and Almuqren, Latifa and Alshahrani, Hadeel Saeed and Aljohani, Raneem Oudah},
  title = {Designing a Novel CNN–LSTM-based Model for Arabic Handwritten Character Recognition for the Visually Impaired Person},
  journal = {Journal of Disability Research},
  year = {2025},
  volume = {4},
  number = {1},
  doi = {10.57197/jdr-2024-0080}
}
Kostenlos starten →