Syamsuddin, Ana Ainul Syamsi (2016) Arabic Optical Character Recognition using Artificial Neural Network Method. Tugas Akhir (S1) - thesis, UNIVERSITAS BAKRIE.
Preview |
Text (PDF)
00 Cover.pdf - Submitted Version Download (1MB) | Preview |
![[thumbnail of pdf]](https://repository.bakrie.ac.id/style/images/fileicons/text.png) |
Text (pdf)
01 BAB I - III.pdf - Submitted Version Restricted to Registered users only Download (756kB) |
|
Text (pdf)
02 BAB IV.pdf - Submitted Version Restricted to Registered users only Download (945kB) |
|
Text (pdf)
03 BAB V.pdf - Submitted Version Restricted to Registered users only Download (91kB) |
|
Text (pdf)
04 DAFTAR PUSTAKA.pdf - Submitted Version Restricted to Registered users only Download (187kB) |
|
Text (pdf)
05 Lampiran.pdf - Submitted Version Restricted to Registered users only Download (1MB) |
Abstract
Optical Character Recognition (OCR) is the process of converting scanned images of machine printed or handwritten text into a computer based format. It involves computer software that designed to translate images of text into machine printed editable text, or to translate pictures of characters into a standard encoding scheme representing them in ASCII or Unicode. This research will focus on OCR for Arabic script with all its unique characteristics. Feed Forward Neural Network Method with Back-propagation algorithm for training and testing stage were used. Using APTI data set, the research conducted with nearly 6000 images of both isolated and cursive characters. The research work in three main stages, pre-processing, feature extraction, and classification or recognition. Pre-processing stage consists of binarization, complement, normalization, and thinning. Segmentation stage also provided. Zoning, 2D DCT, and GLCM were implemented for Feature Extraction stage. Best algorithm that give best result respectively as follows: binarization, complement, segmentation, normalization, thinning, feature extraction, and classification. The proposed method yields the best accuracy rate up to 96.08% for 19 character classes experiment using Zoning method. While accuracy rate for 38 character classes experiment achieved up to 72.43% using 2D DCT method. K-fold cross validation also implemented and increased the accuracy rate for each method. So that, it proven effectively well support method for Artificial Neural Network.
| Item Type: | Thesis (Tugas Akhir (S1) - ) |
|---|---|
| Uncontrolled Keywords: | OCR, Feed Forward Neural Network, Back-propagation algorithm |
| Subjects: | Computer Science Thesis > Thesis (S1) |
| Divisions: | Fakultas Teknik dan Ilmu Komputer > Program Studi Informatika |
| Depositing User: | Ana Ainul Syamsi S |
| Date Deposited: | 15 Sep 2016 04:38 |
| Last Modified: | 15 Sep 2016 04:38 |
| URI: | https://repository.bakrie.ac.id/id/eprint/372 |
Actions (login required)
 |
View Item |