Off-line Arabic handwritten character recognition using neural network /
Character Recognition (CR) is considered as one of the most important field in pattern recognition. The ultimate objectives of the Optical Character Recognition (OCR) system is to simulate the capability of reading, hence OCR can be considered as artificial intelligence. The challenges of the Arabic...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
Kuala Lumpur :
Kulliyyah of Engineering, International Islamic University Malaysia,
2018
|
| Subjects: | |
| Online Access: | Click here to view 1st 24 pages of the thesis. Members can view fulltext at the specified PCs in the library. |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Character Recognition (CR) is considered as one of the most important field in pattern recognition. The ultimate objectives of the Optical Character Recognition (OCR) system is to simulate the capability of reading, hence OCR can be considered as artificial intelligence. The challenges of the Arabic handwriting have several parameters and directives that gives the style uniqueness such as the writing direction, the shape of the characters, etc. In this research, a character handwritten recognition for the Arabic language is developed. The main aim of the system is to save time and effort of Arabic OCR. In addition, it can become alternative of the manual typing due to the reliability and time taken. In the proposed system, data has been collected for 28 Arabic characters which form new dataset to be applied in the system. The system has four main stages; preprocessing, segmentation, feature extraction, classification and recognition. The system is off-line and depends on the image acquisition. So, acquitted image has to go through these four stages. The feed-forwarded Neural Network was used as a classifier. Seven features have been used for the recognition stage which is the blob basic region (bonding box, area and centroid). In addition, it focuses on the character that have similarities and the system will also consider the number of dots and its position, and the connected components. The proposed system was able to recognize as many characters as it can with accuracy rate of 95.23% as compared to the AIA9K dataset which reached 84.96%. However, the recognition system has some limitation such as the difficulty to extract features for the AIA9K due to the size of the image in the dataset. |
|---|---|
| Physical Description: | xii, 75 leaves : colour illustrations ; 30cm. |
| Bibliography: | Includes bibliographical references (leaves 62-66). |