A comparative study of different approaches of primitive printed Arabic Optical Character Recognition

This paper is a part of comprehensive research and development towards eventually enhanced AOCR (Arabic Optical Character Recognition) accuracy. This paper presents a comparative study of four most reported as high primitive printed Arabic Character Recognition works (Hamdi et al. [1], Marwa et al. [2], Marwa et al. [3] and Andrey et al. [4]). The motivation was the lack of independent comparative study between recent algorithms for researchers to depend on. A new noise-free generated dataset of 7 font types is used for evaluation. The results shown that Marwa et al. [3] algorithm overcomes the others across all dataset tests.