To search, Click below search items. |
| All Published Papers Search Service |
|
|
|
Title |
Digital Restoration of Degraded Historical Document of Yoruba Cultural Manuscript Using Generative Adversarial Network and Linear Shade Algorithm |
|
Author |
Abdulquadri, Sharafadeen, Oludare Isaac Abiodun, Okike Benjamin, Abiodun Esther Omolara |
| Citation |
Vol. 26 No. 3 pp. 159-181 |
|
Abstract |
Over the years the preservation and restoration of historical handwritten Yoruba documents is a pressing concern due to their vulnerability to degradation, deterioration, and loss. Despite efforts to address this issue using deep learning, existing methods often fall short in addressing document-specific challenges, such as text clarity. This study proposes the use of a novel approach combining Generative Adversarial Networks (GANs) with the LSHADE algorithm to restore Yoruba documents. The proposed LSHADE-GAN model is designed to overcome the limitations of traditional GANs, including unstable training, sensitive hyper-parameters, and slow convergence. The model's performance was evaluated using five samples and compared to two conventional deep learning models. The results show that LSHADE-GAN outperforms the other models in terms of Peak Signal-to-Noise Ratio (PSNR) and Mean Squared Error (MSE). The LSHADE-GAN model achieved a PSNR of 13.25, significantly higher than DE-GAN and PSO-GAN, which had PSNR values ranging from 5.91 to 12.97. The model also demonstrated a lower MSE of 0.04, compared to DE-GAN (0.05) and PSO-GAN (0.07). Qualitative assessments revealed improved visual quality, including reduced noise, clearer text, and enhanced readability. These results demonstrate the potential of LSHADE-GAN model to achieve adequate restoration of degraded Yoruba documents and significantly improve text clarity. This breakthrough has significant implications for the preservation of cultural heritage, enabling the recovery of valuable historical information and facilitating a deeper understanding of the past. The expected result will be restoration of degraded documents and text clarity. With this expected outcome the conclusion can be drawn as well as state the implication of study with recommendation of future work. |
|
Keywords |
Restoration of historical documents, Restoration of handwritten Yoruba documents, Yoruba cultural manuscript, LSHADE-GAN model. |
|
URL |
