A Comprehensive Approach to Arabic Handwriting Recognition: Deep Convolutional Networks and Bidirectional Recurrent Models for Arabic Scripts | ||
| International Journal of Telecommunications | ||
| Volume 04, Issue 02, July 2024, Pages 1-11 PDF (1.42 M) | ||
| Document Type: Original Article | ||
| DOI: 10.21608/ijt.2024.291347.1052 | ||
| Authors | ||
| Ayman Saber* 1; Ahmed Taha1; Khalid Abd El Salam2 | ||
| 1Electrical Engineering Department, Suez Canal University, Ismailia, Egypt. | ||
| 2Electrical Engineering Department, Suez Canal University, Ismailia, Egypt. Department of Information System, College of Information Technology, Misr University for Science and Technology (MUST), 6th of October City 12566, Egypt. | ||
| Abstract | ||
| Arabic handwriting recognition presents unique challenges due to the complexities of Arabic calligraphy and variations in writing styles. Proposing a novel approach to address these challenges by leveraging advanced deep learning techniques. This focus is on Convolutional Neural Networks (CNNs) and Bidirectional Long Short-Term Memory (Bi-LSTM) networks, which are tailored specifically for recognizing handwritten Arabic text. Utilizing the KHATT dataset for comprehensive training and evaluation, implementing rigorous preprocessing steps to enhance data quality. Central to this methodology is the Res-Net152 architecture for feature extraction, which has proven highly effective. This approach achieved remarkable results, with a character error rate of approximately 2.96% and an accuracy of 97.04% on the testing dataset. These results significantly outperform the previous method, representing a substantial advancement in the field of Arabic handwriting recognition. The study demonstrates the potential of deep learning models in overcoming the unique challenges posed by Arabic script, paving the way for further improvements and applications. | ||
| Keywords | ||
| Optical Character Recognition (OCR); Artificial Neural Networks; Text segmentation; Document Digitization; KHATT Dataset | ||
|
Statistics Article View: 404 PDF Download: 251 |
||
