Skripsi

KLASIFIKASI TULISAN TANGAN ARAB DENGAN HYBRID MOMENT INVARIANT-BACKPROPAGATION (HMI-BP) DAN CONVOLUTIONAL NEURAL NETWORK (CNN)

---

This study presents a comparative analysis of Arabic handwritten character classification using the Hybrid Moment Invariant–Backpropagation (HMI-BP) and Convolutional Neural Network (CNN) approaches. The morphological complexity of Arabic script and the high variability in writing styles among writers demand models capable of distinguishing visually similar characters. The Arabic Handwritten Character Dataset (AHCD), consisting of 13,440 training and 3,360 testing images, was used in this study. The HMI-BP method extracted seven Hu Moment Invariants as invariant features, which were classified using a multilayer perceptron with three hidden layers, achieving an accuracy of 44.82%, precision of 44.47%, recall of 44.82%, and an F1-score of 44.17%. In contrast, the CNN model, which performs end-to-end learning through multiple convolutional layers, achieved significantly better results with an accuracy of 97.92%, precision of 97.95%, recall of 97.92%, and F1-score of 97.91%. These findings demonstrate that CNN is more effective in extracting spatial features and handling variations in Arabic handwriting. The results highlight the superiority of deep learning–based approaches for Arabic handwriting recognition and provide a foundation for developing more accurate OCR systems in the future.

---

Availability

Detail Information

Series Title

-

Call Number

T1899432025

Publisher

Indralaya : Prodi Teknik Informatika, Fakultas Ilmu Komputer Universitas Sriwijaya., 2025

Collation

xv, 139 hlm.; ilus.; tab.; 29 cm

Language

Indonesia

ISBN/ISSN

-

Classification

006.420 7

Content Type

Text

Media Type

-

Carrier Type

-

Edition

-

Subject(s)

Pengenalan Karakter Tulisan Tangan
Prodi Teknik Informatika

Specific Detail Info

-

Statement of Responsibility

MI

Other version/related

File Attachment

Comments

---

You must be logged in to post a comment