Identifikasi Huruf Hijaiyah Berbasis Glcm Menggunakan Jaringan Syaraf Tiruan Backpropagation

Katya Blinda Putri, Noveri Lysbetti Marpaung

Abstract


Hijaiyah Letter or also called the Arabic alphabet can be found in the Holy Book of Islam, the Qur'an. Every Muslim is obliged to read the Qur'an and recognize each of Hijaiyah Letters as its constituent letters in order to read and write in Arabic correctly. Humans are essentially intelligent because they are able to distinguish between one object and another through the object's characteristics or patterns. Certainly it is not difficult to identify a handwriting of Hijaiyah Letters even though written by different people. But it's hard if a machine is trying to identify it. The research aims to establish an Identification System of Hijaiyah Letters based on Gray Level Cooccurence Matrix (GLCM) using the Backpropagation Artificial Neural Network to determine system accuracy results and produce sound output in according to how to read it. Image Data of Hijaiyah Letters obtained from hand-written scanning of five different people with total of Hijaiyah Letters as much as 600 data measuring by 300x300 pixels with extension. JPEG for each letter. GLCM-based systems as the extraction of texture features produce values of Energy, Contrast, Homogeneity, Entropy and Correlation, used as input for the training process of Backpropagation Artificial Neural Network. Image Data of Hijaiyah letters used for the training process amounted to 420 letters and 180 letters of testing process. The result of the Hijaiyah Letters Identification System identifies 30 letters from Alif to Ya. The system is able to identify Hijaiyah Letters with an accuracy rate of 96.11% with error 3.89%.
Keywords: Hijaiyah Letters, GLCM, Backpropagation Artificial Neural Network, Matlab R2018a.

Full Text:

PDF

Refbacks

  • There are currently no refbacks.