To search, Click below search items. |
| All Published Papers Search Service |
|
|
|
Title |
Optical Character Recognition using Deep Learning: An enhanced Approach |
|
Author |
Marwa Amara and Radhia Zaghdoud |
| Citation |
Vol. 22 No. 5 pp. 545-552 |
|
Abstract |
Optical character recognition (OCR) is a critical activity that has a wide range of applications. The most important stage in any system is still the segmentation of the script to be transmitted for the recognition system. Modeling and feature extraction are done with these smaller components that make up the word. As a result, the segmentation stage is thought to be the primary source of recognition mistakes. One of the most serious issues with Arabic character segmentation systems is the absence of research on Arabic character features. In reality, a successful segmentation system necessitates awareness of Arabic topography. To address this problem, we've created multiple roles that contribute to promote the issue. Then, to decide whether to affirm segmentation points or re-do the segmentation, we suggest a binary support vector machine (SVM). Finally, deep learning is used to classify the characters. |
|
Keywords |
OCR; handwritten segmentation; SVM; deep learning |
|
URL |
