A Comparative Study on Arabic Handwritten Words Recognition Using Textures Descriptors

Abstract

Nowadays, the recognition of Arabic handwritten become one of the most challenging issues because of the intrinsic characteristics of the Arabic language. In this work, we investigate the performance of several texture descriptors on the recognition of Arabic handwritten words. We propose to describe Arabic words using Local Binary Patterns (LBP) and Gray Level Co- occurrence Matrix (GLCM). In addition, we propose to use, for the first time, the Multi Level Local Phase Quantization (ML-LPQ) descriptor. To conduct classification, we use three supervised classifiers namely Support Vector Machine (SVM), Naïve Bayes (NB) and K-Nearest Neighbors (KNN). As a second contribution, we introduce a new database of Arabic handwritten that is made up of 1000 words from the computer science field. Experimental evaluation has shown promising results.